A method and system for remote visualization of intersection data

Abstract

The embodiment of the application relates to a kind of intersection data remote visualization processing method and system, the method comprises: cloud platform monitors the equipment online state of each intersection;Real-time monitoring data of each intersection is received;The three-dimensional scene of each intersection is simulated;Real-time analysis is carried out to the traffic event of each intersection;Real-time analysis is carried out to the traffic index of each intersection;Real-time analysis is carried out to the number of traffic participants of each intersection;Real-time traffic condition of each intersection is visualized processing. Through the application, the intelligent analysis level of road intersection can be enhanced, and the real-time analysis capability of visual monitoring can be improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a remote visualization processing method and system for intersection data. Background Technology

[0002] Visual monitoring of traffic conditions at road intersections is a typical application requirement in current traffic management. The conventional solution involves deploying cameras at the front end to capture real-time video, playing the video backend, and having backend staff manually analyze the traffic conditions. Clearly, this conventional solution relies too heavily on manual analysis, and its real-time performance and accuracy need improvement. Summary of the Invention

[0003] The purpose of this invention is to address the shortcomings of existing technologies by providing a remote visualization processing method and system for intersection data. Multiple roadside points are set up at each intersection, and a set of monitoring devices and a roadside point communication device are installed at each roadside point. Each monitoring device connects to a remote cloud platform through its corresponding roadside point communication device. Each monitoring device monitors the traffic conditions of the current intersection and the current roadside point in real time and sends the monitoring data to the cloud platform. The cloud platform, on the one hand, uses an intersection monitoring database to receive and store the real-time monitoring data from the front end; on the other hand, based on an asynchronous processing mechanism, it extracts information from the intersection monitoring database to perform 3D scene construction, traffic event analysis, traffic indicator analysis, and traffic participation value analysis, storing the dynamic analysis results in the intersection analysis database. Furthermore, customized first and third visualization pages display the real-time monitoring video, traffic event analysis, traffic indicator analysis, and traffic participation value analysis of each intersection, and a customized second visualization page uses digital twin technology to display the scene of the current intersection from the perspective of the current roadside point. This invention can enhance the intelligent analysis level of road intersections and improve the real-time analysis capabilities of visual monitoring.

[0004] To achieve the above objectives, a first aspect of the present invention provides a method for remote visualization processing of intersection data, the method comprising:

[0005] The cloud platform monitors the online status of equipment at each intersection and updates the intersection equipment database based on the monitoring results;

[0006] The system receives real-time monitoring data from each intersection and stores it in the intersection monitoring database.

[0007] The three-dimensional scene of each intersection is simulated based on the intersection monitoring database, and the simulation results are stored in the corresponding three-dimensional scene data table in the intersection analysis database;

[0008] The traffic events at each intersection are analyzed in real time based on the intersection monitoring database, and the analysis results are stored in the corresponding traffic event data table in the intersection analysis database.

[0009] The traffic indicators of each intersection are analyzed in real time based on the intersection monitoring database, and the analysis results are stored in the corresponding traffic indicator data table in the intersection analysis database.

[0010] The number of traffic participants at each intersection is analyzed in real time based on the intersection monitoring database, and the analysis results are stored in the corresponding traffic participant data table in the intersection analysis database.

[0011] The real-time traffic conditions at each intersection are visualized based on the real-time updated databases of intersection equipment, intersection monitoring, and intersection analysis, as well as the pre-set databases of operating vehicles and intersection-roadside point relationships.

[0012] Preferably, each intersection corresponds to a unique intersection number, which is recorded as the corresponding first intersection number; each intersection includes multiple intersection branches, and each intersection branch corresponds to a branch direction, which includes east, south, west, and north. If the number of intersection branches with a single direction is not unique, they are sequentially encoded based on the current direction; the motor vehicles traveling on each road at each intersection include both commercial and non-commercial vehicles; each commercial vehicle is pre-installed with an OBU device;

[0013] At each intersection, a roadside point is pre-set on the roadside leading to the corresponding intersection, denoted as the corresponding first roadside point; each first roadside point is assigned a unique roadside point number, denoted as the corresponding first roadside point number; each first roadside point is pre-set with a first roadside point communication device and multiple first monitoring devices.

[0014] The first roadside communication device locally stores the corresponding first intersection number, first roadside number, and first roadside orientation; the first roadside orientation is consistent with the corresponding branch orientation.

[0015] Each of the first monitoring devices is connected to the cloud platform through a corresponding first roadside point communication device; the first monitoring device locally stores a set of corresponding device parameters, including the first device name, the first device number, the first device type, and the first device manufacturer; the first device type includes telephoto cameras, panoramic cameras, lidar, millimeter-wave radar, RSU devices, and traffic signal poles;

[0016] The first monitoring device, which is a telephoto camera, is used to capture real-time video of the monitored road to generate first real-time monitoring data containing a fixed-length video. The latest first real-time monitoring data is periodically sent to the cloud platform through the corresponding first roadside communication device at a preset synchronization frequency. The first real-time monitoring data includes a first timestamp, a first device number, a first data type, and first video data. The first data type is set to telephoto video type.

[0017] The first monitoring device, which is a panoramic camera, is used to capture real-time video of the monitored road to generate a second real-time monitoring data containing a fixed-length video. The latest second real-time monitoring data is periodically sent to the cloud platform through the corresponding first roadside communication device at a preset synchronization frequency. The second real-time monitoring data includes a second timestamp, the first device number, a second data type, and second video data. The second data type is set to panoramic video type.

[0018] The first monitoring device, which is a lidar device, is used to perform radar scanning on the monitoring environment to generate third real-time monitoring data. It periodically sends the latest third real-time monitoring data to the cloud platform via the corresponding first roadside communication device at a preset synchronization frequency. The third real-time monitoring data includes a third timestamp, the first device number, a third data type, and a first radar point cloud. The third data type is set to lidar point cloud type. The features of each point in the first radar point cloud include first coordinate features and first reflection intensity features. The coordinate system of the first coordinate features is the world coordinate system.

[0019] The first monitoring device, which is a millimeter-wave radar, is used to perform radar scanning on the monitoring environment to generate a fourth real-time monitoring data, and periodically sends the latest fourth real-time monitoring data to the cloud platform through the corresponding first roadside point communication device at a preset synchronization frequency; the fourth real-time monitoring data includes a fourth timestamp, the first device number, a fourth data type, and a second radar point cloud; the fourth data type is set to millimeter-wave radar point cloud type; the features of each point in the second radar point cloud include a second coordinate feature and a first velocity feature; the coordinate system of the second coordinate feature is the world coordinate system;

[0020] The first monitoring device, whose first device type is an RSU device, is used to receive first operating vehicle data sent by the OBU devices of each operating vehicle within the monitoring range, and to assemble all the first operating vehicle data received in the most recent first time period into corresponding fifth real-time monitoring data according to a preset first time period length; and to periodically send the latest fifth real-time monitoring data to the cloud platform through the corresponding first roadside point communication device at a preset synchronization frequency; the first operating vehicle data includes a first vehicle timestamp, a first vehicle license plate, a first vehicle model, a first vehicle color, a first driving mode, a first driver identifier, a first operating agency identifier, a first vehicle location, and a first vehicle speed; the first driving mode includes unmanned driving, automatic driving, and manual driving; the first driver identifier is the identity identifier of the current driver when the first driving mode is automatic driving or manual driving; the fifth real-time monitoring data includes a fifth timestamp, the first device number, a fifth data type, and all the first operating vehicle data received in the most recent first time period, and the time interval between the first vehicle timestamps of every two first operating vehicle data in the fifth real-time monitoring data does not exceed the first time period length; the fifth data type is set as the operating vehicle type;

[0021] The first monitoring device, whose first device type is a traffic signal pole, is used to acquire the real-time light status of all traffic lights on the pole, generate corresponding sixth real-time monitoring data, and periodically send the latest sixth real-time monitoring data to the cloud platform through the corresponding first roadside communication device at a preset synchronization frequency. The sixth real-time monitoring data includes a sixth timestamp, the first device number, a sixth data type, and multiple first traffic light data. The first traffic light data includes the first traffic light type, the first traffic light status, and the remaining duration of the first traffic light. The first traffic light type includes left turn light type, straight light type, and right turn light type. The first traffic light status includes red light status, yellow light status, yellow light flashing status, and green light status. The sixth data type is set to the light pole type.

[0022] The first roadside point communication device is used to, when receiving real-time monitoring data sent by any type of the first monitoring device, take the first, second, third, fourth, fifth or sixth real-time monitoring data received at that time as the corresponding current real-time monitoring data, and send the corresponding first roadside point data packet composed of the first intersection number, the first roadside point number, the first roadside point orientation and the current real-time monitoring data to the cloud platform;

[0023] The first roadside point communication device is also used to periodically detect whether the online status of all the first monitoring devices connected to it is normal, obtain the corresponding first device online status list, and send the first device heartbeat command carrying the first intersection number, the first roadside point number, and the first device online status list to the cloud platform; the first device online status list includes multiple first device status records; the first device status record includes a first monitoring device number field and a first monitoring device online status field; the first monitoring device online status field includes online status and offline status.

[0024] Preferably, the cloud platform includes the operating vehicle database, the intersection-roadside point relationship database, the intersection equipment database, the intersection monitoring database, and the intersection analysis database;

[0025] The operating vehicle database includes multiple first vehicle records; each first vehicle record includes a first vehicle identifier field, a first vehicle license plate field, a first vehicle model field, a first vehicle color field, a first driving mode field, a first driver field, and a first operating organization field; the first driving mode field includes driverless, autonomous, and manual driving; the first driver field is empty when the first driving mode field is driverless, and is a designated driver identifier when the first driving mode field is autonomous or manual driving.

[0026] The intersection-roadside point relationship database includes multiple first relationship records; each first relationship record includes a first intersection number field, a first intersection name field, a first intersection center point coordinate field, and a first roadside point set field; the first roadside point set field is used to store the corresponding first roadside point set; the first roadside point set includes multiple first roadside point records; each first roadside point record includes a first roadside point number field, a first roadside point orientation field, and a first roadside point coverage area field.

[0027] The intersection equipment database includes multiple intersection equipment data tables, and each intersection equipment data table corresponds one-to-one with the first intersection number;

[0028] The intersection equipment data table includes multiple first equipment records; each first equipment record includes a second roadside point number field, a first equipment number field, a first equipment name field, a first equipment type field, a first equipment manufacturer field, a first equipment status field, and a first equipment image field; the first equipment type field includes telephoto cameras, panoramic cameras, LiDAR, millimeter-wave radar, RSU equipment, and traffic signal poles; the first equipment status field includes online status and offline status.

[0029] The intersection monitoring database includes multiple intersection monitoring sub-databases, each of which corresponds one-to-one with the first intersection number; each of the intersection monitoring sub-databases includes a video data table, a point cloud data table, an operating vehicle data table, and a traffic light pole data table;

[0030] The video data table includes multiple first video records; each first video record includes a third side point number field, a second device number field, a second device type field, a first timestamp field, and a first video field; the second device type field includes telephoto cameras and panoramic cameras;

[0031] The point cloud data table includes multiple first point cloud records; each first point cloud record includes a fourth side point number field, a third device number field, a third device type field, a second timestamp field, and a first radar point cloud field; the third device type field includes lidar and millimeter-wave radar.

[0032] The operating vehicle data table includes multiple second vehicle records; each second vehicle record includes a fifth roadside point number field, a fourth device number field, a third timestamp field, a second vehicle license plate field, a second vehicle model field, a second vehicle color field, a second driving mode field, a second driver field, a second operating organization field, a first vehicle location field, and a first vehicle speed field; the second driving mode field includes driverless, autonomous, and manual driving; the second driver field is empty when the second driving mode field is driverless, and is the current driver's identity identifier when the second driving mode field is autonomous or manual driving;

[0033] The traffic light pole data table includes multiple first pole records; each first pole record includes a sixth side point number field, a fifth device number field, a fourth timestamp field, and a first traffic light set field; the first traffic light set field is used to store the corresponding first traffic light set; the first traffic light set includes multiple first traffic light records; each first traffic light record includes a first traffic light type field, a first traffic light status field, and a first traffic light remaining duration field; the first traffic light type field includes left turn light, straight light, and right turn light; the first traffic light status field includes red light status, yellow light status, yellow light flashing status, and green light status.

[0034] The intersection analysis database includes multiple intersection analysis sub-databases, each corresponding one-to-one with the first intersection number; each intersection analysis sub-database includes the 3D scene data table, the traffic event data table, the traffic indicator data table, and the traffic participant data table;

[0035] The three-dimensional scene data table includes multiple first scene records; the first scene record includes a seventh side point number field, a fifth timestamp field, and a first intersection three-dimensional scene map field;

[0036] The traffic incident data table includes multiple first incident records; each first incident record includes the eighth roadside point number field, the first incident type field, the first incident location field, the first incident impact range field, the first incident time field, and the first incident evidence video field.

[0037] The traffic indicator data table includes multiple first indicator records; the first indicator record includes the ninth roadside point number field, the fifth timestamp field, the first lane identifier field, the first lane traffic efficiency field, the first lane average speed field, the first lane average delay time field, the first lane average number of stops field, and the first lane average queue length field.

[0038] The traffic participant data table includes multiple first participant records; each first participant record includes a tenth roadside point number field, a sixth timestamp field, a first type of traffic participant quantity field, a second type of traffic participant quantity field, and a third type of traffic participant quantity field; the traffic participant types corresponding to the first, second, and third type of traffic participant quantity fields are people, motor vehicles, and non-motor vehicles, respectively.

[0039] Preferably, the cloud platform monitors the online status of equipment at each intersection and updates the intersection equipment database based on the monitoring results, specifically including:

[0040] The cloud platform assigns a corresponding first timer to each of the first roadside communication devices and times them at a normal time frequency.

[0041] Upon receiving a heartbeat command from a first roadside communication device, the system resets the corresponding first timer and restarts the timing. It extracts the first intersection number, the first roadside point number, and the first device online status list from the heartbeat command as the corresponding current intersection number, current roadside point number, and current device online status list. It also uses the intersection device data table corresponding to the current intersection number as the corresponding current intersection device data table. Furthermore, it iterates through each first device status record in the current device online status list, extracting the first monitoring device number field and the first monitoring device online status field from the currently iterated first device status record as the corresponding current monitoring device number and current monitoring device online status. Finally, it resets the first device status field of the first device record in the current intersection device data table where the second roadside point number field matches the current roadside point number and the first device number field matches the current monitoring device number to the current monitoring device online status.

[0042] The system will identify in real time whether the current timing result of each of the first timers exceeds the preset timing threshold. If it does, the system will take the first roadside point number and the first intersection number corresponding to the first roadside point communication device corresponding to the current first timer as the corresponding current offline roadside point number and current offline intersection number, and reset the first device status field of all the first device records in the intersection device data table corresponding to the current offline intersection number to the offline status.

[0043] Preferably, the step of receiving and storing real-time monitoring data from each intersection in the intersection monitoring database specifically includes:

[0044] When the cloud platform receives the first roadside point data packet, it extracts the first intersection number, the first roadside point number, the first roadside point orientation, and real-time monitoring data as the corresponding current intersection number, current roadside point number, current roadside point orientation, and current real-time monitoring data. It then uses the video data table, point cloud data table, operating vehicle data table, and traffic light pole data table from the intersection monitoring sub-database corresponding to the current intersection number as the corresponding current video data table, current point cloud data table, current operating vehicle data table, and current traffic light pole data table. Furthermore, it extracts the timestamp and device number from the current real-time monitoring data as the corresponding current timestamp and current device number. Finally, it queries the intersection device data table corresponding to the current intersection number, extracts the first device type field from the first device record where the second roadside point number field matches the current roadside point number and the first device number field matches the current device number, and extracts the data type from the current real-time monitoring data as the corresponding current data type.

[0045] When the current data type is telephoto video, the corresponding first video data is extracted from the current real-time monitoring data; and the corresponding first video record is added to the current video data table, consisting of the current roadside point number, the current device number, the current device type, the current timestamp, and the first video data as corresponding fields.

[0046] When the current data type is panoramic video, the corresponding second video data is extracted from the current real-time monitoring data; and the corresponding first video record is added to the current video data table, consisting of the current roadside point number, the current device number, the current device type, the current timestamp, and the second video data as corresponding fields.

[0047] When the current data type is a lidar point cloud, the corresponding first lidar point cloud is extracted from the current real-time monitoring data; and the corresponding first point cloud record is added to the current point cloud data table, consisting of the current roadside point number, the current device number, the current device type, the current timestamp, and the first lidar point cloud as corresponding fields.

[0048] When the current data type is millimeter-wave radar point cloud, the corresponding second radar point cloud is extracted from the current real-time monitoring data; and the corresponding first point cloud record is added to the current point cloud data table, consisting of the current roadside point number, the current device number, the current device type, the current timestamp, and the second radar point cloud as corresponding fields.

[0049] When the current data type is an operating vehicle type, multiple first operating vehicle data are extracted from the current real-time monitoring data; and the corresponding second vehicle record is added to the current operating vehicle data table, using the current roadside point number, the current device number, the current timestamp, and the first vehicle license plate, first vehicle model, first vehicle color, first driving mode, first driver identifier, first operating agency identifier, first vehicle location, and first vehicle speed of each first operating vehicle data as corresponding fields;

[0050] When the current data type is a light pole type, multiple first traffic light data are extracted from the current real-time monitoring data; and the first traffic light type, first traffic light status, and remaining duration of each first traffic light data are used as corresponding fields to form a corresponding first traffic light record, and all the obtained first traffic light records are used to form a corresponding first traffic light set; and the current roadside point number, the current device number, the current timestamp, and the first traffic light set are used as corresponding fields to form a corresponding first light pole record, which is added to the current traffic light pole data table.

[0051] Preferably, the step of simulating the 3D scene of each intersection based on the intersection monitoring database and storing the simulation results in the corresponding 3D scene data table in the intersection analysis database specifically includes:

[0052] When the cloud platform adds a first point cloud record with the third device type field being LiDAR to any of the point cloud data tables, it takes the currently added first point cloud record as the corresponding type of point cloud record, extracts the second timestamp field of the type of point cloud record as the corresponding current timestamp, takes the point cloud data table with the newly added record as the corresponding current point cloud data table, takes the intersection monitoring sub-database and the first intersection number corresponding to the current point cloud data table as the corresponding current intersection monitoring sub-database and current intersection number, takes the video data table of the current intersection monitoring sub-database as the corresponding current video data table, extracts the fourth roadside point number field of the type of point cloud record as the corresponding current roadside point number, and extracts the first intersection center point coordinate field of the first relationship record in the intersection-roadside point relationship database that matches the first intersection number field with the current intersection number as the corresponding current intersection center point coordinate. Then, it performs intersection area map extraction processing from a preset high-precision map with the current intersection center point coordinate as the center to obtain the corresponding current intersection map.

[0053] The first point cloud record in the current point cloud data table whose fourth side point number field is the current roadside point number, whose third device type field is millimeter-wave radar, and whose time interval between the second timestamp field and the current timestamp is less than a set time threshold is extracted as the corresponding second type point cloud record; and the first video record in the current video data table whose third side point number field is the current roadside point number, whose second device type field is panoramic camera, and whose time interval between the first timestamp field and the current timestamp is less than a set time threshold is extracted as the corresponding current panoramic video record;

[0054] If neither the type II point cloud record nor the current panoramic video record is empty, then the first radar point cloud field of the type I and type II point cloud records is extracted as the corresponding type I and type II point clouds; and the first video field of the current panoramic video record is extracted as the corresponding current panoramic video; each point of the type I point cloud corresponds to a world coordinate and a reflection intensity; each point of the type II point cloud corresponds to a world coordinate and a relative velocity.

[0055] Points with non-zero relative velocity in the second type of point cloud are considered moving points. Points in the first type of point cloud corresponding to each moving point and points exceeding the current intersection map coordinate range are deleted. The first type of point cloud with all points deleted is used as the corresponding current point cloud. Single-frame images are extracted from the current panoramic video to obtain multiple first-frame images. The image clarity of each first-frame image is identified, and the image with the highest clarity is selected as the corresponding current scene image.

[0056] A bird's-eye view of the intersection is constructed based on the current intersection map to obtain a corresponding bird's-eye view scene image of the current intersection; multiple 3D first target detection boxes are obtained by performing target detection on the current point cloud based on a point cloud target detection model; multiple first target mask images with depth features are obtained by performing semantic segmentation processing on the current scene image based on a visual image segmentation model with depth estimation; a corresponding first matching group is formed by the first target detection boxes and the first target mask images corresponding to the same target; and the height, appearance, and color of the corresponding target in the current intersection bird's-eye view scene image are reconstructed in 3D based on each of the first matching groups to obtain a corresponding 3D scene image of the current intersection; each pixel of the current intersection 3D scene image inherits the corresponding world coordinates from the current intersection map through the current intersection bird's-eye view scene image;

[0057] The first scene record, composed of the current roadside point number, the current timestamp, and the current intersection 3D scene map as corresponding fields, is added to the 3D scene data table of the intersection analysis sub-database corresponding to the current intersection number.

[0058] Preferably, the step of performing real-time analysis of traffic events at each intersection based on the intersection monitoring database and storing the analysis results in the corresponding traffic event data table within the intersection analysis database specifically includes:

[0059] When the cloud platform adds the first video record with the second device type field being a telephoto camera to any of the video data tables, it takes the currently added first video record as the corresponding current video record, and extracts the three-way side point number field, the second device number field, and the first video field of the current video record as the corresponding current roadside point number, current device number, and current video. It also takes the video data table with the newly added record as the corresponding current video data table, and takes the intersection monitoring sub-database and the first intersection number corresponding to the current video data table as the corresponding current intersection monitoring sub-database and current intersection number.

[0060] Based on a preset event classification model, the current video is processed for event detection and classification to obtain the corresponding event type, event location, event range, and event time. The first video fields of all first video records in the current video data table, where the third roadside point number field matches the current roadside point number, the second device number field matches the current device number, and the first timestamp field is within a specified time range before and after the event time, are extracted and stitched together in chronological order to obtain the corresponding event evidence video. The corresponding first event record, composed of the current roadside point number, event type, event location, event range, event time, and event evidence video as corresponding fields, is added to the traffic event data table of the intersection analysis sub-database corresponding to the current intersection number. The event classification model includes at least a pedestrian violation event analysis model, a non-motorized vehicle violation event analysis model, a motorized vehicle road violation event analysis model, and a motorized vehicle driving violation event analysis model.

[0061] Preferably, the step of performing real-time analysis of traffic indicators at each intersection based on the intersection monitoring database and storing the analysis results in the corresponding traffic indicator data table within the intersection analysis database specifically includes:

[0062] The cloud platform iterates through each of the intersection monitoring sub-databases at a specified time interval. During the iteration, the video data table and traffic light pole data table of the currently iterated intersection monitoring sub-database are used as the corresponding current video data table and current traffic light pole data table. The first intersection number corresponding to the currently iterated intersection monitoring sub-database is used as the corresponding current intersection number, and the traffic indicator data table of the intersection analysis sub-database corresponding to the current intersection number is used as the corresponding current traffic indicator data table.

[0063] All first video records in the current video data table whose first timestamp field is within the most recent first specified time period and whose second device type field is a telephoto camera are extracted to form a corresponding set. The first video records in the set are then clustered by roadside point to obtain multiple first record sets. Similarly, all first light pole records in the current traffic light pole data table whose fourth timestamp field is within the most recent first specified time period are extracted to form a corresponding set. The first light pole records in the set are then clustered by roadside point to obtain multiple second record sets. A corresponding first set group is formed by the first and second record sets corresponding to the same roadside point. Based on the intersection number and roadside point number corresponding to the first set group, a high-precision road map of the corresponding road at the corresponding roadside point location under the corresponding intersection is obtained from a preset high-precision map as the corresponding first road map. All first video records in the first record set have the same third roadside point number field, all first light pole records in the second record set have the same sixth roadside point number field, and all records in the first set group have the same roadside point number field.

[0064] The videos of the first video field of all the first video records in the first record set of the first set group are extracted and spliced ​​together in chronological order to obtain the corresponding first long video; the first long video is processed by single-frame image extraction to obtain multiple second frame images, and a target classification model based on visual images is used to perform vehicle target detection and classification processing on each second frame image to obtain multiple first vehicle target detection boxes to form a corresponding first frame vehicle target set; and a conventional target tracking algorithm is used to perform vehicle target tracking processing on all the first frame vehicle target sets to obtain multiple first vehicle trajectories; and the average speed of the corresponding first vehicle is obtained by dividing the trajectory length of each first vehicle trajectory by the trajectory duration; the first vehicle target detection box includes the detection box target type, detection box center coordinates, detection box size and detection box orientation angle, and the detection box target type is the vehicle type, which includes cars, buses, engineering vehicles, trucks and vans;

[0065] Based on the second record set of the first set group, the time points when various traffic light types switched from other states to green light states within the most recent first specified time period are identified to obtain one or more first time points to form a corresponding first time point sequence;

[0066] The first lane traffic index data set is obtained by estimating the average vehicle speed, traffic efficiency, and average delay time of each lane in the first road map. Specifically, the number of trajectory points of each first vehicle trajectory in each lane is counted to obtain the corresponding first lane point number, and the current first vehicle trajectory is taken as the subordinate trajectory of the lane with the largest number of first lane points; the average speed of the first vehicles in all first vehicle trajectories under each lane is averaged to obtain the corresponding first lane average speed; the traffic efficiency of each lane is calculated to obtain the corresponding first lane traffic efficiency = (first lane average speed / free flow speed) * 100%; and the average delay time of each lane is calculated to obtain the corresponding first lane average delay time = (intersection average length / first lane average speed) - (intersection average length / free flow speed); and the first lane traffic index data set is composed of the lane markings of each lane and the corresponding first lane average speed, first lane traffic efficiency, and first lane average delay time; the free flow speed is a preset fixed speed value, and the intersection average length is a preset fixed length value.

[0067] The average number of stops and average queue length of each lane in the first road map are estimated to obtain the corresponding second lane traffic index data set. Specifically, a fixed vehicle length is assigned to each type of vehicle as the corresponding type vehicle length; each lane is traversed; during traversal, the currently traversed lane is recorded as the corresponding current lane, and the corresponding traffic light type of the current lane is obtained from the first road map and recorded as the current lane traffic light type. Each of the first time points in the first time point sequence corresponding to the current lane traffic light type is recorded as the corresponding second time point, and the number of the second time points is counted to generate the corresponding first total. Two counters starting with 0 and one data sequence starting with empty are initialized for the current lane and recorded as the corresponding first vehicle counter, first stop count counter, and first vehicle length sequence. It is confirmed whether each first vehicle trajectory has intersected with the current lane once or more. If it is confirmed that the trajectory has intersected once or more, the first vehicle counter is incremented. The counter value is incremented by 1; and it is confirmed whether each of the first vehicle trajectories is in the current lane at each of the second time points. If it is confirmed that the current first vehicle trajectory is in the current lane at the current second time point, the counter value of the first parking vehicle counter is incremented by 1, and the vehicle length corresponding to the vehicle type of the current first vehicle trajectory is added to the first vehicle length sequence; after the first vehicle counter and the first parking vehicle counter have completed counting and the first vehicle length sequence has also completed data addition, the corresponding first lane average number of stops is calculated as first parking vehicle counter / first vehicle counter, and the first vehicle length sequence is summed to generate the corresponding first vehicle length sum and the corresponding first lane average queue length is calculated as first vehicle length sum / first total; and at the end of the traversal, the lane identifier of each lane and the corresponding first lane average number of stops and first lane average queue length are combined to form the corresponding second lane traffic indicator data group;

[0068] The current platform time is used as the current timestamp; and the corresponding first indicator record is added to the current traffic indicator data table by using the roadside point number corresponding to each of the first set groups, the current timestamp, the lane identifier corresponding to each lane, the first lane traffic efficiency, the first lane average speed, the first lane average delay time, the first lane average number of stops, and the first lane average queue length as corresponding fields.

[0069] Preferably, the step of performing real-time analysis of the number of traffic participants at each intersection based on the intersection monitoring database and storing the analysis results in the corresponding traffic participant data table within the intersection analysis database specifically includes:

[0070] The cloud platform iterates through each of the intersection monitoring sub-databases at a specified time interval. During the iteration, the video data table of the currently iterated intersection monitoring sub-database is used as the corresponding current video data table, the first intersection number corresponding to the currently iterated intersection monitoring sub-database is used as the corresponding current intersection number, and the traffic participant data table of the intersection analysis sub-database corresponding to the current intersection number is used as the corresponding current traffic participant data table.

[0071] Extract the first roadside point number fields from each of the first relationship records in the intersection-roadside point relationship database that match the first intersection number field with the current intersection number, and use them as the corresponding first numbers; extract the first timestamp field and first video field from the first video record in the current video data table that match the third roadside point number field with each of the first numbers, has the second device type field as a panoramic camera, and has the first timestamp field as the one closest to the current time, and use them as the corresponding first panoramic timestamp and second panoramic video; extract single-frame images from each of the second panoramic videos and use the last frame as the corresponding first panoramic image; and use a visual image-based target classification model to process the first panoramic image. Multiple second target detection boxes are obtained through target detection and classification. The total number of second target detection boxes for target types (people, motor vehicles, and non-motor vehicles) is counted to obtain the corresponding number of first, second, and third traffic participants. The first participant record is added to the current traffic participant data table by using the first number, the corresponding first panoramic timestamp, and the number of first, second, and third traffic participants as corresponding fields. The second target detection box includes target type, target coordinates, target size, and target orientation angle. Target types include people, animals, motor vehicles, non-motor vehicles, and buildings. The second target detection box includes the detection box target type, detection box center coordinates, detection box size, and detection box orientation angle.

[0072] Preferably, the method further includes: pre-setting three visualization pages, namely a first visualization page, a second visualization page, and a third visualization page; wherein,

[0073] The first visualization page includes a first, second, third, and fourth display area. The first display area includes an intersection name entry and multiple intersection number entries. The second display area includes a total number of intersection monitoring devices and multiple entries for the number of first-class devices. The third display area includes a map area, a total number of roadside point monitoring devices, and a list of roadside point monitoring devices. Each record in the roadside point monitoring device list includes the device name, device number, device orientation, and device manufacturer fields. The fourth display area includes a list of roadside point traffic events. Each record in the roadside point traffic event list includes the event type, event location, event range, event time, and evidence viewing fields.

[0074] The second visualization page is a visualization page implemented based on digital twin technology;

[0075] The third visualization page includes a monitoring video area, a first analysis area, and a second analysis area. The monitoring video area includes east, south, west, and north video areas and an intersection twin video area. The intersection twin video area is a video area implemented based on digital twin technology, including orientation entries and comparison markers. The first analysis area includes traffic efficiency entries, average vehicle speed entries, average delay time entries, average number of stops entries, average queue length entries, and an envelope diagram display area. The second analysis area includes a statistical chart display area, pedestrian quantity entries, motor vehicle quantity entries, and non-motor vehicle quantity entries.

[0076] Preferably, the step of visualizing the real-time traffic conditions of each intersection based on the real-time updated intersection equipment database, intersection monitoring database, intersection analysis database, and pre-set operating vehicle database and intersection-roadside point relationship database specifically includes:

[0077] The cloud platform loads the first visualization page;

[0078] The total number of the first relationship records in the intersection-roadside point relationship database is counted to obtain the corresponding total number of first intersections; and an intersection number entry for the total number of first intersections is created in the first display area of ​​the first visualization page. All intersection number entries are browsed by flipping through the triangular page-turning symbols on the left and right sides of the first display area. A one-to-one correspondence is established between each intersection number entry and the first relationship record. The display content of the corresponding intersection number entry is set by the first intersection number field of each first relationship record. After the setting is completed, the first intersection number entry is selected as the currently selected intersection number entry.

[0079] When any of the intersection number entries is selected, the first relationship record corresponding to the currently selected intersection number entry is taken as the corresponding current relationship record, and the first intersection number field of the current relationship record is extracted as the corresponding current intersection number. The display content of the intersection name entry in the first display area is set according to the first intersection name field of the current relationship record.

[0080] In the intersection equipment data table corresponding to the current intersection number, the number of equipment types is statistically analyzed to generate a corresponding first type quantity. The total number of the first equipment records is statistically analyzed to obtain the corresponding first equipment total quantity. The total number of the first equipment records of each type of the same equipment is statistically analyzed to obtain the first type quantity of the first category of equipment. The display content of the intersection monitoring equipment total quantity entry in the second display area is set according to the first equipment total quantity. An entry for the first type of equipment quantity is created in the second display area, and a one-to-one correspondence is established between each entry for the first type of equipment quantity and the total number of the first type of equipment. The corresponding first type of equipment quantity information is composed of each total number of the first type of equipment and the corresponding equipment type name. The display content of the corresponding entry for the first type of equipment quantity is set according to the first type of equipment quantity information.

[0081] Based on the center point coordinate field of the first intersection in the current relationship record, the corresponding high-precision map of the intersection is extracted from the preset high-precision map and loaded into the map area of ​​the third display area; based on the orientation field of the first roadside point records of the first roadside point set in the current relationship record, roadside point marking is drawn on the high-precision map of the intersection in the map area; when any roadside point mark is selected, the currently selected roadside point mark is enlarged to generate the corresponding current roadside point mark; a prompt box is used above the current roadside point mark to provide a prompt description of the current roadside point; and the first roadside point number field corresponding to the current roadside point mark in the current relationship record is extracted as the corresponding current roadside point number;

[0082] In the intersection equipment data table corresponding to the current intersection number, all first equipment records that match the second roadside point number field with the current roadside point number are extracted to form a corresponding first record list; the total number of first equipment records in the first record list is counted to obtain the corresponding second total number of equipment; the display content of the total number of roadside point monitoring equipment entries in the third display area is set based on the second total number of equipment; and the content of each record in the roadside point monitoring equipment list in the third display area is set based on the first record list.

[0083] In the traffic event data table of the intersection analysis sub-database corresponding to the current intersection number, all the first event records that most recently match the ninth roadside point number field with the current roadside point number are extracted to form a corresponding second record list; and based on the first event type field, first event location field, first event impact range field, and first event time field of each record in the second record list, the display content of the event type, event location, event range, and event time fields of each record in the roadside point traffic event list in the fourth display area is set; and a default viewing mark is set in the evidence viewing field of each record in the roadside point traffic event list; and when any of the viewing marks is clicked, the video in the first event evidence viewing video field of the first event record corresponding to the current viewing mark in the second record list is played through a pop-up window;

[0084] The cloud platform loads the second visualization page;

[0085] The first video field, which matches the third roadside point number field with the current roadside point number, has the second device type field being the panoramic camera, and the first timestamp field being the first video record closest to the current time, is extracted from the video data table of the intersection monitoring sub-database corresponding to the current intersection number as the corresponding current panoramic video. The first intersection 3D scene map field, which matches the seventh roadside point number field with the current roadside point number, and has the fifth timestamp field being the first scene record closest to the current time, is extracted from the 3D scene data table of the intersection analysis sub-database corresponding to the current intersection number as the corresponding first intersection 3D scene map.

[0086] The system simulates corresponding roadside equipment poles at various roadside points in the 3D scene of the first intersection using preset roadside equipment pole visualization objects; it also simulates monitoring equipment on each roadside equipment pole visualization object using preset monitoring equipment visualization objects, and sets the icons and names of each monitoring equipment visualization object according to the monitoring equipment information provided by the intersection equipment data table corresponding to the current intersection number; furthermore, it simulates the monitoring coverage area under each roadside equipment pole visualization object using preset roadside point coverage area visualization objects, and sets the size of each roadside point coverage area visualization object according to the coverage area field of each first roadside point in the current relationship record; and uses the resulting 3D scene of the first intersection as the corresponding baseline 3D scene.

[0087] The current panoramic video is processed to extract multiple first-frame panoramic images by extracting single-frame images. Each first-frame panoramic image is then traversed, and during traversal, the currently traversed first-frame panoramic image is used as the corresponding current panoramic image. Semantic segmentation processing is performed on the current panoramic image based on a visual image segmentation model with depth estimation to obtain multiple second target mask images with depth features. The world coordinates of each second target mask image are obtained based on the transformation relationship between image coordinates and world coordinates. In the baseline 3D scene, at the scene location corresponding to the world coordinates of the second target mask image representing each target semantic type (human, animal, motor vehicle, or non-motorized vehicle), a corresponding human, animal, motorized vehicle, or non-motorized vehicle is created. The system visualizes vehicles or non-motorized vehicles and sets their appearance based on the image features of the second target mask image. The set baseline 3D scene is recorded as the corresponding first frame 3D scene. Based on the first vehicle positioning field of all second vehicle records in the operating vehicle data table of the intersection monitoring sub-database corresponding to the current intersection number that are closest to the current panoramic image and have a time interval less than a set threshold, the system identifies whether each visualized motor vehicle in the first frame 3D scene is an operating vehicle. If so, the first vehicle speed field of the corresponding second vehicle record is extracted as the corresponding first target speed, and a preset visualization speed marker is used. The object simulates speed prompt information above the visualized motor vehicle object, and sets the speed prompt information of the visualized speed marker object based on the first target vehicle speed; and creates a traffic light sign visualized object above the scene position corresponding to the world coordinates of the second target mask image with the semantic type of traffic light in the first frame 3D scene, and sets the display content of the traffic light color, traffic light type and remaining traffic light duration of each corresponding traffic light sign visualized object based on the first light pole record at each roadside point in the traffic light pole data table of the intersection monitoring sub-database corresponding to the current intersection number, which is closest to the time of the current panoramic image; and in the In the first frame of the 3D scene, a traffic information sign visualization object is created above each of the traffic light sign visualization objects. Based on the intersection high-precision map and the traffic indicator data table of the intersection analysis sub-database corresponding to the current intersection number, the first indicator record at each roadside point that is closest to the time of the current panoramic image is used to set the display content of lane traffic direction, lane flow rate, and lane queue length for each of the traffic information sign visualization objects. At the end of the traversal, all the obtained first frame 3D scenes are sorted in chronological order to obtain the corresponding first frame 3D scene sequence. Digital twin video conversion is performed on the first frame 3D scene sequence to obtain the corresponding first twin video.Each of the second target mask images corresponds to a target semantic type, and the target semantic type includes people, animals, motor vehicles, non-motor vehicles, and traffic lights;

[0088] The second visualization page loads and plays the first twin video; if the user selects any of the monitoring device visualization objects during playback, a device sign visualization object is created on the selected monitoring device visualization object, and the device sign visualization object displays the name, online status, device image, type, number, orientation, and manufacturer of the current monitoring device according to the current relationship record and the intersection device data table corresponding to the current intersection number;

[0089] The cloud platform loads the third visualization page;

[0090] Then, from the video data table of the intersection monitoring sub-database corresponding to the current intersection number, the first video field of the first video record with the second device number field being a telephoto camera, which is the latest in time and the time interval between each roadside point does not exceed a preset time threshold, is extracted as the corresponding first roadside point real-time video; and according to the roadside point orientation corresponding to each first roadside point real-time video, the corresponding video areas in the four video areas of east, south, west and north within the monitoring video area are loaded and played; and after any of the four video areas is selected, the corresponding video area is loaded and played. The orientation is set as the current orientation, and the real-time video of the first roadside point currently playing in the current video area is set as the corresponding real-time video of the current roadside point. The current roadside point number is modified to match the roadside point number corresponding to the current video area. In the intersection twin video area of ​​the monitoring video area, the real-time video of the current roadside point is rendered using digital twin technology. The orientation entry in the intersection twin video area is set to the corresponding current orientation. When the comparison mark in the intersection twin video area is clicked, the real-time video of the current roadside point and the corresponding rendered video are switched and compared in the intersection twin video area.

[0091] Then, from the traffic indicator data table of the intersection analysis sub-database corresponding to the current intersection number, the first lane traffic efficiency field, the first lane average speed field, the first lane average delay time field, the first lane average number of stops field, and the first lane average queue length field of each lane corresponding to the current roadside point number are extracted as the corresponding first traffic efficiency, first average speed, first average delay time, first average number of stops field, and first average queue length; and the first traffic efficiency, first average speed, first average delay time, and first average number of stops field for all lanes are calculated. The average values ​​of the first roadside point traffic efficiency, the first roadside point average vehicle speed, the first roadside point average delay time, the first roadside point average number of stops, and the first roadside point average queue length are calculated by averaging the first average queue length. Based on the first roadside point traffic efficiency, the first roadside point average vehicle speed, the first roadside point average delay time, the first roadside point average number of stops, and the first roadside point average queue length, the display content of the traffic efficiency item, the average vehicle speed item, the average delay time item, the average number of stops item, and the average queue length item in the first analysis area are set.

[0092] When the traffic efficiency item is selected, the traffic efficiency of the first roadside point at each time point within the most recent first specified time period is calculated based on the historical data of the current traffic indicator data table to obtain the corresponding first roadside point traffic efficiency sequence. A first envelope diagram is then plotted in the envelope diagram display area with time as the horizontal axis and efficiency percentage as the vertical axis based on the first roadside point traffic efficiency sequence. When the average vehicle speed item is selected, the average vehicle speed of the first roadside point at each time point within the most recent first specified time period is calculated based on the historical data of the current traffic indicator data table to obtain the corresponding first roadside point average vehicle speed sequence. A second envelope diagram is then plotted in the envelope diagram display area with time as the horizontal axis and vehicle speed as the vertical axis based on the first roadside point average vehicle speed sequence. When the average delay time item is selected, the average delay time of the first roadside point at each time point within the most recent first specified time period is calculated based on the historical data of the current traffic indicator data table to obtain the corresponding first roadside point average delay time. The system calculates the average delay time series of the first roadside points and plots a third envelope diagram in the envelope diagram display area with time as the horizontal axis and delay time as the vertical axis. When the average number of stops is selected, it calculates the average number of stops at the first roadside points at each time point within the most recent first specified time period based on the historical data of the current traffic indicator data table. Based on this first roadside point average number of stops sequence, it plots a fourth envelope diagram in the envelope diagram display area with time as the horizontal axis and number of stops as the vertical axis. When the average queue length is selected, it calculates the average queue length at the first roadside points at each time point within the most recent first specified time period based on the historical data of the current traffic indicator data table. Based on this first roadside point average queue length sequence, it plots a fifth envelope diagram in the envelope diagram display area with time as the horizontal axis and queue length as the vertical axis.

[0093] Then, from the traffic participant data table of the intersection analysis sub-database corresponding to the current intersection number, the number fields of the first, second, and third types of traffic participants that match the current intersection number in the tenth roadside point number field and whose sixth timestamp field is the most recent time are extracted as the corresponding first pedestrian number, first motor vehicle number, and first non-motor vehicle number; and the display content of the pedestrian number entry, motor vehicle number entry, and non-motor vehicle number entry in the second analysis area is set based on the first pedestrian number, first motor vehicle number, and first non-motor vehicle number.

[0094] When the pedestrian number item, the motor vehicle number item, or the non-motor vehicle number item is selected, the number of the first pedestrian, the number of the first motor vehicle, or the number of the first non-motor vehicle at each time point within the most recent second specified time period is obtained based on the historical data of the current traffic participant data table, thereby forming the corresponding first pedestrian number sequence, first motor vehicle number sequence, or first non-motor vehicle number sequence. Based on the first pedestrian number sequence, the first motor vehicle number sequence, or the first non-motor vehicle number sequence, the corresponding first, second, or third curve is plotted in the statistical chart display area with time as the horizontal axis and quantity as the vertical axis.

[0095] The cloud platform refreshes the content of the first, second, and third visualization pages at a preset refresh frequency.

[0096] A second aspect of this invention provides a system for implementing the remote visualization processing method for intersection data described in the first aspect above. The system includes: a cloud platform, multiple first roadside point communication devices, and multiple first monitoring devices; each of the first monitoring devices is connected to the cloud platform through a corresponding first roadside point communication device.

[0097] The cloud platform is used to monitor the online status of equipment at each intersection and update the intersection equipment database based on the monitoring results; receive real-time monitoring data from each intersection and store it in the intersection monitoring database; simulate the 3D scene of each intersection based on the intersection monitoring database and store the simulation results in the corresponding 3D scene data table in the intersection analysis database; perform real-time analysis of traffic events at each intersection based on the intersection monitoring database and store the analysis results in the corresponding traffic event data table in the intersection analysis database; perform real-time analysis of traffic indicators at each intersection based on the intersection monitoring database and store the analysis results in the corresponding traffic indicator data table in the intersection analysis database; perform real-time analysis of the number of traffic participants at each intersection based on the intersection monitoring database and store the analysis results in the corresponding traffic participant data table in the intersection analysis database; and visualize the real-time traffic conditions of each intersection based on the real-time updated intersection equipment database, intersection monitoring database, intersection analysis database, and the pre-set operating vehicle database and intersection-roadside point relationship database.

[0098] Preferably, each intersection includes multiple intersection branches, and a roadside point is pre-set on the roadside of each intersection branch leading to the corresponding intersection, denoted as the corresponding first roadside point; a first roadside point communication device and multiple first monitoring devices are pre-set on each first roadside point.

[0099] This invention provides a remote visualization processing method and system for intersection data. Multiple roadside points are set up at each intersection, and a set of monitoring devices and a roadside point communication device are installed at each roadside point. Each monitoring device connects to a remote cloud platform through its corresponding roadside point communication device. Each monitoring device monitors the traffic conditions of the current intersection and the current roadside point in real time and sends the monitoring data to the cloud platform. The cloud platform receives and stores the real-time monitoring data from the front end using an intersection monitoring database. Simultaneously, based on an asynchronous processing mechanism, it extracts information from the intersection monitoring database to perform 3D scene construction, traffic event analysis, traffic indicator analysis, and traffic participation value analysis, storing the dynamic analysis results in the intersection analysis database. Furthermore, it uses customized first and third visualization pages to display the real-time monitoring video, traffic event analysis, traffic indicator analysis, and traffic participation value analysis of each intersection, and uses a customized second visualization page based on digital twin technology to display the scene of the current intersection from the perspective of the current roadside point. This invention enhances the intelligent analysis level of road intersections and improves the real-time analysis capability of visual monitoring. Attached Figure Description

[0100] Figure 1 This is a schematic diagram of a remote visualization processing method for intersection data provided in Embodiment 1 of the present invention;

[0101] Figure 2 A page structure diagram of the first visual page provided in Embodiment 1 of the present invention;

[0102] Figure 3 This is a schematic diagram of the second visual page provided in Embodiment 1 of the present invention;

[0103] Figure 4 This is a page structure diagram of the third visual page provided in Embodiment 1 of the present invention;

[0104] Figure 5 This is a module structure diagram of a remote visualization processing system for intersection data provided in Embodiment 2 of the present invention. Detailed Implementation

[0105] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0106] This invention provides a remote visualization processing method for intersection data to achieve visualized monitoring of traffic conditions at road intersections. The technical solution of this invention includes implementation schemes for roadside monitoring devices at each intersection and a remote cloud platform. The implementation schemes for the roadside monitoring devices at each intersection will be described first, followed by the implementation scheme for the remote cloud platform.

[0107] (a) The implementation schemes for the side point equipment at each intersection include:

[0108] Each intersection is assigned a unique intersection number, designated as the first intersection number. Each intersection includes multiple branch intersections, each branch intersection having a corresponding direction, including east, south, west, and north. If the number of branch intersections with a single direction is not unique, they are sequentially coded based on the current direction (e.g., East 1, East 2, etc.). Motor vehicles traveling on each road at each intersection include both commercial and non-commercial vehicles. Each commercial vehicle is equipped with an Onboard Unit (OBU) device, which is used to communicate with the Road Side Unit (RSU) devices at the roadside points.

[0109] At each intersection, a roadside point is pre-set on the roadside leading to the corresponding intersection, denoted as the corresponding first roadside point; each first roadside point is assigned a unique roadside point number, denoted as the corresponding first roadside point number; each first roadside point is pre-set with a first roadside point communication device and multiple first monitoring devices; here, under normal circumstances, a roadside point device pole is installed at the roadside point to load the first roadside point communication device and all first monitoring devices corresponding to the current roadside point;

[0110] The first-path side-point communication device locally stores the corresponding first-path intersection number, first-path side-point number, and first-path side-point orientation; the orientation of the first-path side-point is consistent with the orientation of the corresponding branch.

[0111] Each first monitoring device connects to the cloud platform via a corresponding first roadside point communication device. Each first monitoring device locally stores a set of corresponding device parameters, including the first device name, first device number, first device type, and first device manufacturer. First device types include telephoto cameras, panoramic cameras, LiDAR, millimeter-wave radar, RSU devices, and traffic signal poles. Here, the number of first monitoring devices with the first device type of telephoto camera can be one, two, or even more. If there is only one, its lens direction is by default outward from the center of the intersection, used to capture road video of the current roadside point's entry / exit lanes. If there are two or more... When there are two or more cameras, one must be a front-facing camera and one is a rear-facing camera. The front-facing camera's lens is directed outwards from the center of the intersection to capture video of the road's entry / exit lanes at the current roadside point. The rear-facing camera's lens is directed in the opposite direction. The video feeds from both cameras are synchronized by default and can be stitched together by the cloud platform. The first monitoring device, a panoramic camera, typically has a 360° field of view and is used to capture panoramic video of the surrounding environment at the current roadside point. The first monitoring device, whether a lidar or millimeter-wave radar, is used to capture video according to its respective set field of view. The radar (FOV) scans the surrounding environment and / or road conditions of the current roadside point. The cloud platform in this embodiment of the invention constructs a 3D scene based on the scanned point cloud from the radar device. A first monitoring device, of type RSU (Roadside Unit), receives real-time data (location, speed) sent by the OBU (On-Board Unit) of any operating vehicle passing through the current roadside point and transmits it back to the cloud platform, enabling the cloud platform to easily obtain the real-time location and speed of each operating vehicle. A first monitoring device, of type traffic light pole, monitors the real-time status of traffic lights at the current roadside point (signal light type, real-time illumination of various types of lights). The system acquires and transmits the status of traffic lights (including the remaining time of each type of traffic light) to the cloud platform. The first roadside point communication device is used to complete the network of all first monitoring devices at the current roadside point and is responsible for data forwarding between the first monitoring devices and the cloud platform. This network structure of the present invention has strong scalability. Users can load more types of first monitoring devices under the first monitoring devices at each roadside point to collect richer real-time information and transmit it to the cloud platform for analysis, and can also load more first monitoring devices at each roadside point to build more monitoring device networks.

[0112] The first monitoring device, which is a telephoto camera, is used to capture real-time video of the monitored road and generate first real-time monitoring data containing a fixed-length video. The latest first real-time monitoring data is periodically sent to the cloud platform through the corresponding first side-point communication device at a preset synchronization frequency. The first real-time monitoring data includes a first timestamp, a first device number, a first data type, and first video data. The first data type is set to telephoto video type.

[0113] The first monitoring device, a panoramic camera, is used to capture real-time video of the monitored road to generate a second real-time monitoring data containing a fixed-length video. The latest second real-time monitoring data is periodically sent to the cloud platform through the corresponding first side-point communication device at a preset synchronization frequency. The second real-time monitoring data includes a second timestamp, a first device number, a second data type, and second video data. The second data type is set to panoramic video.

[0114] The first monitoring device, a lidar-type device, is used to perform radar scanning on the monitoring environment to generate third real-time monitoring data. It periodically transmits the latest third real-time monitoring data to the cloud platform via a corresponding first-path side-point communication device at a preset synchronization frequency. The third real-time monitoring data includes a third timestamp, a first device number, a third data type, and a first radar point cloud. The third data type is set to lidar point cloud type. The features of each point in the first radar point cloud include first coordinate features and first reflection intensity features. The coordinate system of the first coordinate features is the world coordinate system.

[0115] The first monitoring device, a millimeter-wave radar, is used to perform radar scanning on the monitoring environment to generate a fourth real-time monitoring data. It periodically transmits the latest fourth real-time monitoring data to the cloud platform via a corresponding first-path side-point communication device at a preset synchronization frequency. The fourth real-time monitoring data includes a fourth timestamp, a first device number, a fourth data type, and a second radar point cloud. The fourth data type is set to millimeter-wave radar point cloud. The features of each point in the second radar point cloud include second coordinate features and first velocity features. The coordinate system of the second coordinate features is the world coordinate system.

[0116] The first monitoring device, whose first device type is RSU (Roadside Unit), is used to receive first operating vehicle data sent by the OBU (On-Board Unit) devices of each operating vehicle within the monitoring range. It then combines all the first operating vehicle data received within the most recent first time period into corresponding fifth real-time monitoring data according to a preset first time period length. The latest fifth real-time monitoring data is periodically sent to the cloud platform via the corresponding first side-point communication device at a preset synchronization frequency. The first operating vehicle data includes a first vehicle timestamp, first vehicle license plate, first vehicle model, first vehicle color, first driving mode, first driver identifier, first operating organization identifier, first vehicle location, and first vehicle speed. The first driving mode includes driverless, autonomous, and manual driving. The first driver identifier is the current driver's identity identifier when the first driving mode is autonomous or manual. The fifth real-time monitoring data includes a fifth timestamp, a first device number, a fifth data type, and all first operating vehicle data received within the most recent first time period. The time interval between the first vehicle timestamps of any two first operating vehicle data points within the fifth real-time monitoring data does not exceed the first time period length. The fifth data type is set to the operating vehicle type.

[0117] The first device type is a traffic signal pole monitoring device used to acquire the real-time light status of all traffic lights on the pole, generate corresponding sixth real-time monitoring data, and periodically send the latest sixth real-time monitoring data to the cloud platform through the corresponding first roadside communication device at a preset synchronization frequency; the sixth real-time monitoring data includes a sixth timestamp, a first device number, a sixth data type, and multiple first traffic light data; the first traffic light data includes the first traffic light type, the first traffic light status, and the remaining duration of the first traffic light; the first traffic light type includes left turn light type, straight light type, and right turn light type; the first traffic light status includes red light status, yellow light status, yellow light flashing status, and green light status; the sixth data type is set to the light pole type;

[0118] The first-path side-point communication device is used to, upon receiving real-time monitoring data from any type of first monitoring device, take the first, second, third, fourth, fifth, or sixth real-time monitoring data received at that time as the corresponding current real-time monitoring data, and send the corresponding first-path side-point data packet composed of the first intersection number, the first-path side-point number, the first-path side-point orientation, and the current real-time monitoring data to the cloud platform; in addition, the first-path side-point communication device can also be used to merge the most recently received first, second, third, fourth, fifth, or sixth real-time monitoring data into a single current real-time monitoring data at a fixed period, and send the corresponding first-path side-point data packet composed of the first intersection number, the first-path side-point number, the first-path side-point orientation, and the current real-time monitoring data to the cloud platform.

[0119] The first-path side-point communication device is also used to periodically detect whether the online status of all connected first monitoring devices is normal, obtain the corresponding first device online status list, and send the first device heartbeat command carrying the first intersection number, the first-path side-point number, and the first device online status list to the cloud platform; the first device online status list includes multiple first device status records; the first device status record includes a first monitoring device number field and a first monitoring device online status field; the first monitoring device online status field includes online status and offline status; here, the first-path side-point communication device is specifically used to send a preset online status detection command to each first monitoring device when detecting whether the online status of all connected first monitoring devices is normal. If no return data is received from the current first monitoring device within a specified reception time, the first monitoring device online status field corresponding to the current first monitoring device is set to offline status. If return data is received from the current first monitoring device within a specified reception time but the return data indicates that the device is not working properly, the first monitoring device online status field corresponding to the current first monitoring device is set to offline status. If return data is received from the current first monitoring device within a specified reception time and the return data indicates that the device is working properly, the first monitoring device online status field corresponding to the current first monitoring device is set to online status.

[0120] (II) Implementation Scheme of Remote Cloud Platform

[0121] The remote cloud platform solution in this invention's embodiment consists of two parts: database definition and cloud platform implementation steps. The cloud platform in this invention's embodiment defines at least the following databases: an operating vehicle database, an intersection-roadside point relationship database, an intersection equipment database, an intersection monitoring database, and an intersection analysis database. The database structure will be described first, followed by the cloud platform implementation steps. It should be noted that the database definition given in this invention's embodiment is a logical function definition based on data functionality. In practical applications, data deletion, addition, merging, combination, mapping, and other operations can be performed on any one or more data items / fields / records / tables / databases in this logical function definition based on actual needs. Furthermore, different types of entity data files, databases, or database servers can be selected for implementation based on the query and update frequency characteristics of each data item / field / record / table / database.

[0122] The cloud platform of the method in this embodiment of the invention includes an operating vehicle database, an intersection-roadside point relationship database, an intersection equipment database, an intersection monitoring database, and an intersection analysis database; wherein,

[0123] The operational vehicle database includes multiple first vehicle records; each first vehicle record includes a first vehicle identifier field, a first vehicle license plate field, a first vehicle model field, a first vehicle color field, a first driving mode field, a first driver field, and a first operating organization field; the first driving mode field includes driverless, autonomous, and manual driving; the first driver field is empty when the first driving mode field is driverless, and is a designated driver identifier when the first driving mode field is autonomous or manual driving; here, the operational vehicle database is a pre-built database used to store all known operational vehicle information; under normal circumstances, the operational vehicle database will not be dynamically updated, and data addition, deletion, and update operations will only be performed when operational vehicle information is added, deleted, or updated;

[0124] The intersection-roadside point relationship database includes multiple first relationship records. Each first relationship record includes a first intersection number field, a first intersection name field, a first intersection center point coordinate field, and a first roadside point set field. The first roadside point set field stores the corresponding first roadside point set. Each first roadside point set includes multiple first roadside point records. Each first roadside point record includes a first roadside point number field, a first roadside point orientation field, and a first roadside point coverage field. Here, the intersection-roadside point relationship database is a pre-built database used to store information on all roadside points under all known intersections. The coverage information stored in the first roadside point coverage field can be the sum of the effective monitoring ranges of all monitoring devices on the corresponding roadside point, or it can be the effective monitoring range of one or more designated cameras used to monitor road traffic conditions. Under normal circumstances, the intersection-roadside point relationship database will not be dynamically updated; data addition, deletion, and update operations will only be performed when adding, deleting, and updating intersection-roadside point relationships.

[0125] The intersection equipment database includes multiple intersection equipment data tables, each corresponding one-to-one with the first intersection number. Each intersection equipment data table contains multiple first equipment records. Each first equipment record includes fields for second roadside point number, first equipment number, first equipment name, first equipment type, first equipment manufacturer, first equipment status, and first equipment image. The first equipment type field includes telephoto cameras, panoramic cameras, LiDAR, millimeter-wave radar, RSU equipment, and traffic signal poles. The first equipment status field includes online and offline status. Here, the intersection equipment database is a pre-built database used to store all known roadside point equipment at all intersections. Under normal circumstances, most fields in each record of the intersection equipment database will not be dynamically updated; only the first equipment status field will be dynamically updated. Furthermore, when adding, deleting, or updating intersection equipment, the database record itself or other fields within the record will be added, deleted, or updated.

[0126] The intersection monitoring database includes multiple intersection monitoring sub-databases, each corresponding one-to-one with the first intersection number. Each intersection monitoring sub-database includes video data tables, point cloud data tables, operating vehicle data tables, and traffic light pole data tables. All data tables in all sub-databases within the intersection monitoring database are used to store real-time monitoring data from various monitoring devices and are dynamically updated. Therefore, the intersection monitoring database is a database that is always dynamically updated.

[0127] The video data table includes multiple first video records; each first video record includes a third-path side point number field, a second device number field, a second device type field, a first timestamp field, and a first video field; the second device type field includes telephoto cameras and panoramic cameras;

[0128] The point cloud data table includes multiple first point cloud records; each first point cloud record includes a fourth side point number field, a third device number field, a third device type field, a second timestamp field, and a first radar point cloud field; the third device type field includes lidar and millimeter-wave radar.

[0129] The operational vehicle data table includes multiple second vehicle records; each second vehicle record includes the fifth roadside point number field, the fourth device number field, the third timestamp field, the second vehicle license plate field, the second vehicle model field, the second vehicle color field, the second driving mode field, the second driver field, the second operating organization field, the first vehicle location field, and the first vehicle speed field; the second driving mode field includes driverless, autonomous, and manual driving; the second driver field is empty when the second driving mode field is driverless, and displays the current driver's identity when the second driving mode field is autonomous or manual driving;

[0130] The traffic light pole data table includes multiple first light pole records; each first light pole record includes a sixth side point number field, a fifth device number field, a fourth timestamp field, and a first traffic light set field; the first traffic light set field is used to store the corresponding first traffic light set; each first traffic light set includes multiple first traffic light records; each first traffic light record includes a first traffic light type field, a first traffic light status field, and a first traffic light remaining duration field; the first traffic light type field includes left turn light, straight light, and right turn light; the first traffic light status field includes red light status, yellow light status, yellow light flashing status, and green light status; here, in addition to the above three types, other traffic light types can be added based on actual application scenario requirements, such as U-turn light, left turn + straight light, U-turn + left turn + straight light, right turn + straight light, etc.

[0131] The intersection analysis database comprises multiple intersection analysis sub-databases, each corresponding one-to-one with the first intersection number. Each intersection analysis sub-database includes a 3D scene data table, a traffic event data table, a traffic indicator data table, and a traffic participant data table. Here, all the data tables in all the sub-databases within the intersection analysis database are used to store various real-time analysis results from the cloud platform and are dynamically updated. Therefore, the intersection analysis database is also a database that is always dynamically updated.

[0132] The 3D scene data table includes multiple first scene records; each first scene record includes the seventh side point number field, the fifth timestamp field, and the first intersection 3D scene map field.

[0133] The traffic incident data table includes multiple first incident records; each first incident record includes the Eighth Road Side Point Number field, the First Incident Type field, the First Incident Location field, the First Incident Impact Range field, the First Incident Time field, and the First Incident Evidence Video field.

[0134] The traffic indicator data table includes multiple first indicator records; the first indicator records include the Ninth Road Side Point Number field, the Fifth Timestamp field, the First Lane Identifier field, the First Lane Traffic Efficiency field, the First Lane Average Speed ​​field, the First Lane Average Delay Time field, the First Lane Average Number of Stops field, and the First Lane Average Queue Length field.

[0135] The traffic participant data table includes multiple first participant records; each first participant record includes the 10th roadside point number field, the 6th timestamp field, the number of first-class traffic participants field, the number of second-class traffic participants field, and the number of third-class traffic participants field; the traffic participant types corresponding to the first, second, and third-class traffic participant number fields are people, motor vehicles, and non-motor vehicles, respectively.

[0136] The database structure has been introduced above. The following section will introduce the implementation steps of the cloud platform.

[0137] The implementation steps of the remote cloud platform solution in the method of this invention are as follows: Figure 1 The schematic diagram shows a remote visualization processing method for intersection data provided in Embodiment 1 of the present invention. This method mainly includes the following steps:

[0138] Step 1: The cloud platform monitors the online status of equipment at each intersection and updates the intersection equipment database based on the monitoring results;

[0139] Specifically, this includes: Step 11, the cloud platform assigns a corresponding first timer to each first-path side-point communication device and times it according to the normal time frequency;

[0140] Step 12: Upon receiving a first device heartbeat command from a first side-point communication device, the corresponding first timer is reset and restarted. The first intersection number, first side-point number, and first device online status list are extracted from the first device heartbeat command as the corresponding current intersection number, current side-point number, and current device online status list. The intersection device data table corresponding to the current intersection number is used as the corresponding current intersection device data table. Each first device status record in the current device online status list is traversed. During the traversal, the first monitoring device number field and the first monitoring device online status field of the currently traversed first device status record are extracted as the corresponding current monitoring device number and current monitoring device online status. The first device status field of the first device record in the current intersection device data table that matches the current side-point number field and the current side-point number field and the current monitoring device number field is reset to the current monitoring device online status.

[0141] Step 13: Real-time verification of whether the current timing result of each first timer exceeds the preset timing threshold. If it exceeds the threshold, the first side point number and the first intersection number corresponding to the first side point communication device corresponding to the current first timer are used as the corresponding current offline side point number and current offline intersection number. The first device status field of all first device records in the intersection device data table corresponding to the current offline intersection number that match the second side point number field of the current offline intersection number is reset to offline status.

[0142] Step 2: Receive and store the real-time monitoring data of each intersection into the intersection monitoring database;

[0143] Specifically, this includes: Step 21, when the cloud platform receives the first roadside point data packet, it extracts the first intersection number, the first roadside point number, the first roadside point orientation, and real-time monitoring data from it as the corresponding current intersection number, current roadside point number, current roadside point orientation, and current real-time monitoring data; and it uses the video data table, point cloud data table, operating vehicle data table, and traffic light pole data table of the intersection monitoring sub-database corresponding to the current intersection number as the corresponding current video data table, current point cloud data table, current operating vehicle data table, and current traffic light pole data table; it extracts the timestamp and device number of the current real-time monitoring data as the corresponding current timestamp and current device number; it queries the intersection device data table corresponding to the current intersection number, extracts the first device type field of the first device record in the intersection device data table that matches the second roadside point number field with the current roadside point number and the first device number field with the current device number as the corresponding current device type; and it extracts the data type of the current real-time monitoring data as the corresponding current data type;

[0144] Step 22: When the current data type is telephoto video, extract the corresponding first video data from the current real-time monitoring data; and add the corresponding first video record to the current video data table, using the current roadside point number, current device number, current device type, current timestamp, and first video data as corresponding fields.

[0145] Step 23: When the current data type is panoramic video, extract the corresponding second video data from the current real-time monitoring data; and add the corresponding first video record to the current video data table, using the current roadside point number, current device number, current device type, current timestamp, and second video data as corresponding fields.

[0146] Step 24: When the current data type is LiDAR point cloud, extract the corresponding first radar point cloud from the current real-time monitoring data; and add the corresponding first point cloud record to the current point cloud data table, using the current roadside point number, current device number, current device type, current timestamp, and first radar point cloud as corresponding fields.

[0147] Step 25: When the current data type is millimeter-wave radar point cloud, extract the corresponding second radar point cloud from the current real-time monitoring data; and add the corresponding first point cloud record to the current point cloud data table, using the current roadside point number, current device number, current device type, current timestamp, and second radar point cloud as corresponding fields.

[0148] Step 26: When the current data type is the operating vehicle type, extract multiple first operating vehicle data from the current real-time monitoring data; and add the corresponding second vehicle records to the current operating vehicle data table by using the current roadside point number, current device number, current timestamp, and the first vehicle license plate, first vehicle model, first vehicle color, first driving mode, first driver identifier, first operating agency identifier, first vehicle location, and first vehicle speed of each first operating vehicle data as corresponding fields.

[0149] Step 27: When the current data type is light pole type, extract multiple first traffic light data from the current real-time monitoring data; and use the first traffic light type, first traffic light status, and remaining duration of each first traffic light data as corresponding fields to form the corresponding first traffic light record, and use all the obtained first traffic light records to form the corresponding first traffic light set; and add the corresponding first light pole record to the current traffic light pole data table using the current roadside point number, current device number, current timestamp, and first traffic light set as corresponding fields.

[0150] Step 3: Simulate the 3D scene of each intersection based on the intersection monitoring database and store the simulation results in the corresponding 3D scene data table in the intersection analysis database;

[0151] Specifically, this includes: Step 31, when the cloud platform adds a first point cloud record with the third device type field being LiDAR to any point cloud data table, the currently added first point cloud record is taken as the corresponding type of point cloud record, the second timestamp field of the type of point cloud record is extracted as the corresponding current timestamp, the point cloud data table with the newly added record is taken as the corresponding current point cloud data table, the intersection monitoring sub-database and the first intersection number corresponding to the current point cloud data table are taken as the corresponding current intersection monitoring sub-database and the current intersection number, the video data table of the current intersection monitoring sub-database is taken as the corresponding current video data table, the fourth roadside point number field of the type of point cloud record is extracted as the corresponding current roadside point number; the first intersection center point coordinate field of the first relationship record in the intersection-roadside point relationship database that matches the current intersection number is extracted as the corresponding current intersection center point coordinate, and the intersection area map is extracted from the preset high-precision map with the current intersection center point coordinate as the center to obtain the corresponding current intersection map;

[0152] Step 32: Extract the first point cloud record from the current point cloud data table whose fourth path point number field is the current path point number, whose third device type field is millimeter-wave radar, and whose second timestamp field has a time interval of less than a set time threshold, as the corresponding second type point cloud record; and extract the first video record from the current video data table whose third path point number field is the current path point number, whose second device type field is panoramic camera, and whose first timestamp field has a time interval of less than a set time threshold, as the corresponding current panoramic video record;

[0153] Step 33: If neither the Type II point cloud record nor the current panoramic video record is empty, then extract the first radar point cloud field of the Type I and Type II point cloud records as the corresponding Type I and Type II point clouds; and extract the first video field of the current panoramic video record as the corresponding current panoramic video.

[0154] In this case, each point in a type I point cloud corresponds to a world coordinate and a reflection intensity; each point in a type II point cloud corresponds to a world coordinate and a relative velocity.

[0155] Here, the first type of point cloud is the point cloud generated by lidar scanning. Based on the publicly available lidar working mechanism and the corresponding lidar point cloud characteristics, it can be known that each point in this type of point cloud corresponds to at least a three-dimensional world coordinate and a laser reflection intensity. The second type of point cloud is the point cloud generated by millimeter-wave radar scanning. Based on the publicly available millimeter-wave radar working mechanism and the corresponding millimeter-wave radar point cloud characteristics, it can be known that each point in this type of point cloud corresponds to at least a two-dimensional or three-dimensional world coordinate and a relative velocity.

[0156] Step 34: Points with non-zero relative velocity in the second type of point cloud are considered moving points. Points in the first type of point cloud that correspond to each moving point and points that exceed the current intersection map coordinate range are deleted. The first type of point cloud with the points deleted is used as the corresponding current point cloud. Single-frame images are extracted from the current panoramic video to obtain multiple first-frame images. The image clarity of each first-frame image is identified, and the image with the highest clarity is selected as the corresponding current scene image.

[0157] Here, based on the publicly available lidar point cloud features, it is known that the accuracy of target shape confirmation based on this type of point cloud is relatively high. However, this type of point cloud cannot distinguish the motion characteristics of the object where the scan point is located. Therefore, the method of this embodiment eliminates the corresponding points in the first type of point cloud based on the moving points in the second type of point cloud. This can eliminate the interference of the dynamic object point cloud as much as possible while retaining all static object point clouds. In addition, based on the publicly available lidar point cloud features, it is also known that the sparsity of this type of point cloud will increase dramatically beyond a certain range, and the point cloud density will drop sharply, which will greatly affect the accuracy of target shape confirmation. Therefore, the method of this embodiment will also delete points in the first type of point cloud that exceed the coordinate range of the current intersection map. This can greatly reduce the prediction error caused by the fluctuation of point cloud sparsity. It should be noted that when identifying the image clarity of each first frame image, this embodiment can be based on a variety of clarity algorithms. These various clarity algorithms include, but are not limited to, the Tenengrad gradient method, the Laplacian gradient method, and the variance method.

[0158] Step 35: Construct a bird's-eye view of the intersection from the current intersection map to obtain a corresponding bird's-eye view scene image of the current intersection; perform target detection on the current point cloud based on a point cloud target detection model to obtain multiple 3D first target detection boxes; perform semantic segmentation on the current scene image based on a visual image segmentation model with depth estimation to obtain multiple first target mask images with depth features; form a corresponding first matching group with the first target detection boxes and first target mask images corresponding to the same target; and perform 3D reconstruction of the height, appearance, and color of the corresponding target in the current intersection bird's-eye view scene image based on each first matching group to obtain the corresponding 3D scene image of the current intersection; each pixel of the current intersection 3D scene image inherits the corresponding world coordinates from the current intersection map through the current intersection bird's-eye view scene image;

[0159] Here, constructing a Bird's Eye View (BEV) intersection plan view based on a high-precision vector map is a common approach. Many BEV-based AI models (such as BEVDet, BEVFormer, and PersFormer) provide implementation details and references, which will not be repeated here. The intersection bird's eye view and 3D intersection plan view in this embodiment are essentially related to the intersection map... Figure 1 All of these are vector graphics that allow for the addition of data objects and data features. The purpose of constructing the 3D scene map of the intersection using the method of this embodiment is to provide a relatively ideal basic scene framework without dynamic objects and closest to reality for the subsequent digital twin visualization processing steps. The point cloud object detection model used in the method of this embodiment is an artificial intelligence model that can perform object detection and classification on point clouds. Common models include PointNet and SSD, which will not be repeated here. However, it is necessary to customize and train it according to the detection range and classification range before use. The image segmentation models based on visual images used in the method of this embodiment are all artificial intelligence models that can perform pixel-level semantic type segmentation on visual images. Common models include MASK-CNN, which will not be repeated here. However, based on the different segmentation requirements of each step, a customized model version structure needs to be copied in advance based on the general model structure of the image segmentation model, and each customized model version structure needs to be trained separately based on the different classification requirements of each step. After the training is mature, it can be applied.

[0160] Step 36: The first scene record, composed of the current roadside point number, the current timestamp, and the current intersection 3D scene map as corresponding fields, is added to the 3D scene data table of the intersection analysis sub-database corresponding to the current intersection number.

[0161] Step 4: Analyze traffic events at each intersection in real time based on the intersection monitoring database and store the analysis results in the corresponding traffic event data table within the intersection analysis database;

[0162] Specifically, this includes: Step 41, when the cloud platform adds a first video record with the second device type field being a telephoto camera to any video data table, the currently added first video record is taken as the corresponding current video record, and the three-way side point number field, the second device number field, and the first video field of the current video record are extracted as the corresponding current roadside point number, current device number, and current video, and the video data table with the newly added record is taken as the corresponding current video data table, and the intersection monitoring sub-database and the first intersection number corresponding to the current video data table are taken as the corresponding current intersection monitoring sub-database and current intersection number;

[0163] Step 42: Based on the preset event classification model, perform event detection and classification processing on the current video to obtain the corresponding event type, event location, event range, and event time; extract the first video field of all first video records in the current video data table where the third roadside point number field matches the current roadside point number, the second device number field matches the current device number, and the first timestamp field is within a specified time range before and after the event time, and stitch the videos in chronological order to obtain the corresponding event evidence video; and add the corresponding first event record to the traffic event data table of the intersection analysis sub-database corresponding to the current intersection number, using the current roadside point number, event type, event location, event range, event time, and event evidence video as corresponding fields.

[0164] Among them, the event classification model includes at least a pedestrian violation event analysis model, a non-motorized vehicle violation event analysis model, a motorized vehicle road violation event analysis model, and a motorized vehicle driving violation event analysis model.

[0165] Here, the specified time range before and after the event time mentioned in the current step is generally the range from 15 seconds before the event time to 15 seconds after the event time.

[0166] Step 5: Analyze the traffic indicators of each intersection in real time based on the intersection monitoring database and store the analysis results in the corresponding traffic indicator data table in the intersection analysis database.

[0167] Specifically, it includes: Step 51, the cloud platform traverses each intersection monitoring sub-database at a specified time interval, and during the traversal, the video data table and traffic light pole data table of the currently traversed intersection monitoring sub-database are used as the corresponding current video data table and current traffic light pole data table, the first intersection number corresponding to the currently traversed intersection monitoring sub-database is used as the corresponding current intersection number, and the traffic indicator data table of the intersection analysis sub-database corresponding to the current intersection number is used as the corresponding current traffic indicator data table;

[0168] Step 52: Extract all first video records from the current video data table whose first timestamp field is within the most recent first specified duration and whose second device type field is a telephoto camera, form corresponding sets, and cluster the first video records in the sets according to roadside points to obtain multiple first record sets; extract all first light pole records from the current traffic light pole data table whose fourth timestamp field is within the most recent first specified duration, form corresponding sets, and cluster the first light pole records in the sets according to roadside points to obtain multiple second record sets; form a corresponding first set group from the first and second record sets corresponding to the same roadside point; and obtain the high-precision road map of the corresponding road at the corresponding roadside point location under the corresponding intersection from the preset high-precision map based on the intersection number and roadside point number corresponding to the first set group as the corresponding first road map;

[0169] Among them, the third side point number field of all first video records in the first record set is the same, the sixth side point number field of all first light pole records in the second record set is the same, and the roadside point number field of all records in the first set group is the same.

[0170] Step 53: Extract the video from the first video field of all the first video records in the first record set of the first set group and stitch the videos together in chronological order to obtain the corresponding first long video; perform single-frame image extraction processing on the first long video to obtain multiple second frame images, and use a visual image-based target classification model to perform vehicle target detection and classification processing on each second frame image to obtain multiple first vehicle target detection boxes to form the corresponding first frame vehicle target set; and perform vehicle target tracking processing based on a conventional target tracking algorithm according to all the first frame vehicle target sets to obtain multiple first vehicle trajectories; and divide the trajectory length of each first vehicle trajectory by the trajectory duration to obtain the corresponding first vehicle average speed.

[0171] The first vehicle target detection box includes the target type of the detection box, the center coordinates of the detection box, the size of the detection box, and the orientation angle of the detection box. The target type of the detection box is the vehicle type, which includes cars, buses, engineering vehicles, trucks, and vans.

[0172] Here, the target classification models used in the methods of this invention are all artificial intelligence models for target detection and classification based on visual images. Common examples include R-CNN and the YOLO series, which will not be repeated here. However, based on the different classification requirements of each step, a customized model structure needs to be copied in advance based on the general model structure of the target classification model. Each customized model structure is trained separately based on the different classification requirements of each step and then applied after it has matured. The target tracking algorithms used in the methods of this invention are also conventional target tracking algorithms based on the target detection results of visual images at continuous time points, i.e., target detection boxes (bboxes). Examples include the Hungarian algorithm based on the intersection of target detection boxes at different time points, and tracking algorithms based on target state prediction using filters (such as Kalman filters), etc., which will not be repeated here.

[0173] Step 54: Based on the second record set of the first set group, identify the time points when various traffic light types switch from other states to green light states within the most recent first specified time period to obtain one or more first time points to form a corresponding first time point sequence;

[0174] Step 55: Estimate the average vehicle speed, traffic efficiency, and average delay time of each lane in the first road map to obtain the corresponding first lane traffic index data set, specifically:

[0175] The number of trajectory points of each first vehicle trajectory in each lane is counted to obtain the corresponding first lane point count, and the current first vehicle trajectory is taken as the subordinate trajectory of the lane with the largest number of first lane points. The average speed of the first vehicles in all first vehicle trajectories under each lane is averaged to obtain the corresponding first lane average speed. The traffic efficiency of each lane is calculated as (first lane average speed / free flow speed) * 100%. The average delay time of each lane is calculated as (intersection average length / first lane average speed) - (intersection average length / free flow speed). The corresponding first lane traffic indicator data group is composed of lane markings and the corresponding first lane average speed, first lane traffic efficiency, and first lane average delay time.

[0176] The free-flow speed is a preset fixed speed value (e.g., 50 km / h), and the average length of the intersection is a preset fixed length value (e.g., 200 m).

[0177] Step 56: Estimate the average number of stops and average queue length for each lane in the first road map to obtain the corresponding traffic indicator data set for the second lane, specifically:

[0178] A fixed vehicle length is assigned to each vehicle type as the corresponding type vehicle length; each lane is traversed; during traversal, the currently traversed lane is recorded as the current lane, and the corresponding traffic light type of the current lane is obtained from the first road map and recorded as the current lane traffic light type. Each first time point in the first time point sequence corresponding to the current lane traffic light type is recorded as the corresponding second time point, and the number of second time points is counted to generate the corresponding first total. Two counters starting at 0 and one data sequence starting at empty are initialized for the current lane, recorded as the corresponding first vehicle counter, first stop vehicle counter, and first vehicle length sequence. It is confirmed whether each first vehicle trajectory has intersected with the current lane once or more. If it is confirmed that one or more trajectory intersections have occurred, the count value of the first vehicle counter is incremented by 1. The system also tracks each first vehicle trajectory in... The system confirms whether the vehicle is currently in the current lane at each second time point. If it is confirmed that the current first vehicle trajectory is in the current lane at the current second time point, the count value of the first parking vehicle counter is incremented by 1, and the vehicle length corresponding to the vehicle type of the current first vehicle trajectory is added to the first vehicle length sequence. After the first vehicle counter and the first parking vehicle counter have completed counting and the first vehicle length sequence has also completed data addition, the system calculates the corresponding first lane average number of stops = first parking vehicle counter / first vehicle counter, and sums all types of vehicle lengths in the first vehicle length sequence to generate the corresponding first vehicle length sum and calculates the corresponding first lane average queue length = first vehicle length sum / first total. At the end of the traversal, the system forms the corresponding second lane traffic indicator data group by combining the lane identifier of each lane and the corresponding first lane average number of stops and first lane average queue length.

[0179] Step 57: Use the current platform time as the current timestamp; and add the corresponding first indicator record to the current traffic indicator data table, which is composed of the roadside point number corresponding to each first set group, the current timestamp, and the lane identification, first lane traffic efficiency, first lane average speed, first lane average delay time, first lane average number of stops, and first lane average queue length as corresponding fields.

[0180] Step 6: Analyze the number of traffic participants at each intersection in real time based on the intersection monitoring database and store the analysis results in the corresponding traffic participant data table in the intersection analysis database;

[0181] Specifically, it includes: Step 61, the cloud platform traverses each intersection monitoring sub-database at a specified time interval, and during the traversal, the video data table of the currently traversed intersection monitoring sub-database is used as the corresponding current video data table, the first intersection number corresponding to the currently traversed intersection monitoring sub-database is used as the corresponding current intersection number, and the traffic participant data table of the intersection analysis sub-database corresponding to the current intersection number is used as the corresponding current traffic participant data table.

[0182] Step 62: Extract the first roadside point number fields from the first relationship record in the intersection-roadside point relationship database that match the first intersection number field with the current intersection number, and use them as the corresponding first number; extract the first timestamp field and first video field from the first video record in the current video data table that match the third roadside point number field with each first number, have the second device type field of panoramic camera, and have the first timestamp field of the most recent time, and use them as the corresponding first panoramic timestamp and second panoramic video; extract single-frame images from each second panoramic video and use the last frame as the corresponding first panoramic image; use a visual image-based target classification model to perform target detection and classification processing on the first panoramic image to obtain multiple second target detection boxes, and separately count the total number of second target detection boxes for target types of people, motor vehicles, and non-motor vehicles to obtain the corresponding number of first, second, and third type traffic participants; and add the corresponding first participant record to the current traffic participant data table by using each first number, the corresponding first panoramic timestamp, and the number of first, second, and third type traffic participants as corresponding fields.

[0183] The second target detection box includes target type, target coordinates, target size and target orientation angle. Target types include people, animals, motor vehicles, non-motor vehicles and buildings. The second target detection box also includes the target type of the detection box, the center coordinates of the detection box, the size of the detection box and the orientation angle of the detection box.

[0184] Step 7: Visualize the real-time traffic conditions of each intersection based on the real-time updated intersection equipment database, intersection monitoring database, intersection analysis database, as well as the pre-set operating vehicle database and intersection-roadside point relationship database.

[0185] Specifically, this includes: Step 71, the cloud platform loads the first visualization page;

[0186] Here, as Figure 2As shown in the page structure diagram of the first visualization page provided in Embodiment 1 of the present invention, the page display area of ​​the first visualization page includes a first, second, third, and fourth display area; the first display area includes an intersection name entry and multiple intersection number entries; the second display area includes a total number of intersection monitoring devices and multiple first-class device quantity entries; the third display area includes a map area, a total number of roadside point monitoring devices, and a roadside point monitoring device list; each record in the roadside point monitoring device list includes a device name, device number, device orientation, and device manufacturer field; the fourth display area includes a roadside point traffic event list; each record in the roadside point traffic event list includes an event type, event location, event range, event time, and evidence viewing field;

[0187] Step 72: Count the total number of the first relationship records in the intersection-roadside point relationship database to obtain the corresponding total number of the first intersections; create an intersection number entry for the total number of the first intersections in the first display area of ​​the first visualization page, and browse all intersection number entries by flipping through the triangular page-turning symbols on the left and right sides of the first display area, and establish a one-to-one correspondence between each intersection number entry and the first relationship record, and set the display content of the corresponding intersection number entry by the first intersection number field of each first relationship record, and after the setting is completed, first set the first intersection number entry as the currently selected intersection number entry;

[0188] Step 73: When any intersection number entry is selected, the first relationship record corresponding to the currently selected intersection number entry is taken as the corresponding current relationship record, and the first intersection number field of the current relationship record is extracted as the corresponding current intersection number. The display content of the intersection name entry in the first display area is set according to the first intersection name field of the current relationship record.

[0189] Step 74: In the intersection equipment data table corresponding to the current intersection number, the types and quantities of equipment are counted to generate the corresponding first type quantity, and the total number of first equipment records is counted to obtain the corresponding first equipment total quantity. The total number of first equipment records of the same type is counted to obtain the first type quantity of first category equipment total quantity. The display content of the intersection monitoring equipment total quantity entry in the second display area is set according to the first equipment total quantity. The first type quantity of first category equipment is created in the second display area, and a one-to-one correspondence is established between each first type quantity entry and the first type equipment total quantity. The first type equipment quantity information is composed of each first type equipment total quantity and the corresponding equipment type name. The display content of the corresponding first type equipment quantity entry is set according to the first type equipment quantity information.

[0190] Step 75: Extract the corresponding intersection high-precision map from the preset high-precision map based on the first intersection center point coordinate field of the current relationship record and load it into the map area of ​​the third display area; and perform roadside point marking and drawing processing on the intersection high-precision map of the map area based on the first roadside point orientation field of each first roadside point record of the first roadside point set of the current relationship record; when any roadside point mark is selected, perform mark magnification processing on the currently selected roadside point mark to generate the corresponding current roadside point mark; and provide a prompt description of the current roadside point above the current roadside point mark through a prompt box; and extract the first roadside point number field corresponding to the current roadside point mark in the current relationship record as the corresponding current roadside point number;

[0191] Step 76: In the intersection equipment data table corresponding to the current intersection number, extract all first equipment records that match the current roadside point number field to form a corresponding first record list; and calculate the total number of first equipment records in the first record list to obtain the corresponding total number of second equipment; and set the display content of the total number of roadside point monitoring equipment entries in the third display area based on the total number of second equipment; and set the content of each record in the roadside point monitoring equipment list in the third display area based on the first record list.

[0192] Step 77: In the traffic event data table of the intersection analysis sub-database corresponding to the current intersection number, extract all first event records that match the current roadside point number field of the most recent specified number of ninth roadside point numbers to form a corresponding second record list; and set the display content of the event type, event location, event range, and event time fields of each record in the roadside point traffic event list in the fourth display area based on the first event type field, first event location field, first event impact range field, and first event time field of each record in the second record list; and set a default viewing mark in the evidence viewing field of each record in the roadside point traffic event list; and when any viewing mark is clicked, play the video in the first event evidence video field of the first event record corresponding to the current viewing mark in the second record list through a pop-up window;

[0193] Step 78: The cloud platform loads the second visualization page; here, the second visualization page is a visualization page implemented based on digital twin technology.

[0194] Step 79: Extract the first video field from the video data table of the intersection monitoring sub-database corresponding to the current intersection number. The first video field whose third roadside point number matches the current roadside point number, whose second device type field is panoramic camera, and whose first timestamp field is the first video record closest to the current time is extracted as the corresponding current panoramic video. Extract the first intersection 3D scene map field from the 3D scene data table of the intersection analysis sub-database corresponding to the current intersection number. The first intersection 3D scene map field whose seventh roadside point number matches the current roadside point number, and whose fifth timestamp field is the first scene record closest to the current time is extracted as the corresponding first intersection 3D scene map.

[0195] Step 80: Simulate the corresponding roadside equipment poles at each roadside point in the 3D scene of the first intersection using preset roadside equipment pole visualization objects; simulate monitoring equipment on each roadside equipment pole visualization object using preset monitoring equipment visualization objects, and set the icons and names of each monitoring equipment visualization object according to the monitoring equipment information of each roadside point provided by the intersection equipment data table corresponding to the current intersection number; simulate the monitoring coverage area under each roadside equipment pole visualization object using preset roadside point coverage area visualization objects, and set the size of each roadside point coverage area visualization object according to the coverage area field of each first roadside point in the current relationship record; and use the 3D scene of the first intersection obtained after setting as the corresponding baseline 3D scene;

[0196] Step 81: Extract multiple first-frame panoramic images from the current panoramic video using single-frame images; iterate through each first-frame panoramic image; during the iteration, use the currently iterated first-frame panoramic image as the corresponding current panoramic image; perform semantic segmentation on the current panoramic image based on a visual image segmentation model with depth estimation to obtain multiple second target mask images with depth features; obtain the world coordinates of each second target mask image based on the transformation relationship between image coordinates and world coordinates; and create a scene location in the baseline 3D scene corresponding to the world coordinates of the second target mask image with the semantic type of each target (human, animal, motor vehicle, or non-motor vehicle). The system visualizes the corresponding human, animal, motor vehicle, or non-motor vehicle object, and sets the appearance of the current visualized object based on the image features of the second target mask image. The completed baseline 3D scene is recorded as the corresponding first frame 3D scene. Based on the first vehicle positioning field of all second vehicle records in the operating vehicle data table of the intersection monitoring sub-database corresponding to the current intersection number that are closest to the current panoramic image and whose time interval is less than a set threshold, the system identifies whether each motor vehicle visualized object in the first frame 3D scene is an operating vehicle. If so, the first vehicle speed field of the corresponding second vehicle record is extracted as the corresponding first target speed, and this is then used via a preset... The visualized speed marker object simulates speed prompt information above the visualized vehicle object, and sets the speed prompt information of the visualized speed marker object based on the first target vehicle speed; and creates a traffic light sign visualized object above the scene position corresponding to the world coordinates of the mask image of the second target whose semantic type is traffic light in the first frame of the 3D scene, and sets the display content of the traffic light color, traffic light type and remaining traffic light duration of each corresponding traffic light sign visualized object based on the first light pole record of each roadside point that is closest to the current panoramic image time in the traffic light pole data table of the intersection monitoring sub-database corresponding to the current intersection number; and In the first frame of the 3D scene, a traffic information sign visualization object is created above each traffic light sign visualization object. Based on the intersection high-precision map and the traffic indicator data table of the intersection analysis sub-database corresponding to the current intersection number, the first indicator record at each roadside point that is closest to the current panoramic image time is set to display the lane traffic direction, lane flow rate, and lane queue length of each traffic information sign visualization object. At the end of the traversal, all the obtained first frame 3D scenes are sorted in chronological order to obtain the corresponding first frame 3D scene sequence. The first frame 3D scene sequence is then converted into a digital twin video to obtain the corresponding first twin video.

[0197] Each second target mask map corresponds to a target semantic type, which includes people, animals, motor vehicles, non-motor vehicles, and traffic lights.

[0198] Step 82: The second visualization page loads and plays the first twin video. If the user selects any monitoring device visualization object during playback, a device sign visualization object is created on the selected monitoring device visualization object. The device sign visualization object then displays the name, online status, device image, type, number, orientation, and manufacturer of the current monitoring device based on the current relationship record and the intersection device data table corresponding to the current intersection number.

[0199] To provide a clearer understanding of steps 79-82 above, the embodiments of the present invention are as follows: Figure 3 The following is a specific example illustrated in the schematic diagram of the second visualization page provided in Embodiment 1 of the present invention;

[0200] Step 83: The cloud platform loads the third visualization page;

[0201] Here, as Figure 4 As shown in the page structure diagram of the third visualization page provided in Embodiment 1 of the present invention, the page display area of ​​the third visualization page includes a monitoring video area, a first analysis area, and a second analysis area; the monitoring video area includes east, south, west, and north video areas and an intersection twin video area; the intersection twin video area is a video area implemented based on digital twin technology, including orientation entries and comparison marks; the first analysis area includes traffic efficiency entries, average vehicle speed entries, average delay time entries, average number of stops entries, average queue length entries, and an envelope diagram display area; the second analysis area includes a statistical chart display area, pedestrian quantity entries, motor vehicle quantity entries, and non-motor vehicle quantity entries;

[0202] Step 84: Extract the first video field of the first video record of the telephoto camera from the video data table of the intersection monitoring sub-database corresponding to the current intersection number, where the second device number field is the latest time and the time interval between each roadside point does not exceed a preset time threshold. This will be used as the corresponding real-time video of the first roadside point. Then, based on the orientation of the roadside point corresponding to each first roadside point's real-time video, load and play the corresponding video areas in the four video areas (east, south, west, and north) within the monitoring video area. After any one of the four video areas is selected, the current... The orientation of the current video area is taken as the current orientation, and the real-time video of the first roadside point currently playing in the current video area is taken as the corresponding real-time video of the current roadside point. The current roadside point number is modified to the roadside point number corresponding to the current video area. In the intersection twin video area of ​​the monitoring video area, the real-time video of the current roadside point is rendered using digital twin technology. The orientation entry of the intersection twin video area is set to the corresponding current orientation. When the comparison mark in the intersection twin video area is clicked, the real-time video of the current roadside point and the corresponding rendered video are switched and compared in the intersection twin video area.

[0203] Step 85: Extract the first lane traffic efficiency, first lane average speed, first lane average delay time, first lane average number of stops, and first lane average queue length fields of each lane corresponding to the current roadside point number from the traffic indicator data table of the intersection analysis sub-database corresponding to the current intersection number. These fields serve as the corresponding first traffic efficiency, first average speed, first average delay time, first average number of stops, and first average queue length. Then, calculate the first traffic efficiency, first average speed, first average delay time, and first average number of stops for all lanes. The average number of vehicles and the average queue length are calculated to obtain the corresponding first-roadside point traffic efficiency, first-roadside point average vehicle speed, first-roadside point average delay time, first-roadside point average number of stops, and first-roadside point average queue length. Based on the first-roadside point traffic efficiency, first-roadside point average vehicle speed, first-roadside point average delay time, first-roadside point average number of stops, and first-roadside point average queue length, the display content of the traffic efficiency, average vehicle speed, average delay time, average number of stops, and average queue length items in the first analysis area is set.

[0204] Step 86: When the traffic efficiency item is selected, calculate the traffic efficiency sequence of the first roadside points at each time point within the most recent first specified time period based on historical data from the current traffic indicator data table. Then, plot the corresponding first envelope diagram in the envelope diagram display area with time as the horizontal axis and efficiency percentage as the vertical axis based on the first roadside point traffic efficiency sequence. When the average vehicle speed item is selected, calculate the corresponding average vehicle speed sequence of the first roadside points at each time point within the most recent first specified time period based on historical data from the current traffic indicator data table. Then, plot the corresponding second envelope diagram in the envelope diagram display area with time as the horizontal axis and vehicle speed as the vertical axis based on the first roadside point average vehicle speed sequence. When the average delay time item is selected, calculate the corresponding first roadside delay time for each time point within the most recent first specified time period based on historical data from the current traffic indicator data table. The average delay time series of the first-route side points is calculated, and a third envelope diagram is plotted in the envelope diagram display area with time as the horizontal axis and delay time as the vertical axis based on the average delay time series of the first-route side points. When the average number of stops is selected, the average number of stops at the first-route side points is calculated based on the historical data of the current traffic indicator data table for each time point in the most recent first specified time period to obtain the corresponding average number of stops sequence of the first-route side points. A fourth envelope diagram is plotted in the envelope diagram display area with time as the horizontal axis and number of stops as the vertical axis based on the average number of stops sequence of the first-route side points. When the average queue length is selected, the average queue length at the first-route side points is calculated based on the historical data of the current traffic indicator data table for each time point in the most recent first specified time period to obtain the corresponding average queue length sequence of the first-route side points. A fifth envelope diagram is plotted in the envelope diagram display area with time as the horizontal axis and queue length as the vertical axis based on the average queue length sequence of the first-route side points.

[0205] Step 87: Extract the number fields of the first, second, and third types of traffic participants from the traffic participant data table of the intersection analysis sub-database corresponding to the current intersection number. These fields match the current intersection number in the tenth roadside point number field and have the sixth timestamp field as the most recent time. These numbers are then used as the corresponding first pedestrian number, first motor vehicle number, and first non-motor vehicle number. Based on the first pedestrian number, first motor vehicle number, and first non-motor vehicle number, the display content of the pedestrian number, motor vehicle number, and non-motor vehicle number entries in the second analysis area is set.

[0206] Step 88: When the pedestrian number item, motor vehicle number item, or non-motor vehicle number item is selected, the number of first pedestrians, first motor vehicles, or first non-motor vehicles at each time point in the most recent second specified time period is obtained based on the historical data of the current traffic participant data table, so as to form the corresponding first pedestrian number sequence, first motor vehicle number sequence, or first non-motor vehicle number sequence, and the corresponding first, second, or third curves are drawn in the statistical chart display area with time as the horizontal axis and quantity as the vertical axis based on the first pedestrian number sequence, first motor vehicle number sequence, or first non-motor vehicle number sequence;

[0207] Step 89: The cloud platform refreshes the content of the first, second, and third visualization pages according to the preset refresh frequency.

[0208] Figure 5 This is a module structure diagram of a remote visualization processing system for intersection data provided in Embodiment 2 of the present invention. This system can be a terminal device, server, system, or platform implementing Embodiment 1 of the aforementioned method, or it can be a device that enables the aforementioned terminal device, server, system, or platform to implement Embodiment 1 of the aforementioned method. For example, the device can be a device or chip system of the aforementioned terminal device, server, system, or platform. Figure 5 As shown, the system includes: a cloud platform 110, multiple first-side-point communication devices 111, and multiple first-monitoring devices 112; each first-monitoring device 112 is connected to the cloud platform 110 through its corresponding first-side-point communication device 111.

[0209] The cloud platform 110 is used to monitor the online status of equipment at each intersection and update the intersection equipment database based on the monitoring results; it also receives real-time monitoring data from each intersection and stores it in the intersection monitoring database; it simulates the 3D scene of each intersection based on the intersection monitoring database and stores the simulation results in the corresponding 3D scene data table in the intersection analysis database; it performs real-time analysis of traffic events at each intersection based on the intersection monitoring database and stores the analysis results in the corresponding traffic event data table in the intersection analysis database; it performs real-time analysis of traffic indicators at each intersection based on the intersection monitoring database and stores the analysis results in the corresponding traffic indicator data table in the intersection analysis database; it performs real-time analysis of the number of traffic participants at each intersection based on the intersection monitoring database and stores the analysis results in the corresponding traffic participant data table in the intersection analysis database; and it visualizes the real-time traffic conditions at each intersection based on the real-time updated intersection equipment database, intersection monitoring database, intersection analysis database, as well as the pre-set operating vehicle database and intersection-roadside point relationship database. It should be noted that each intersection includes multiple intersection branches. A roadside point is pre-set on the roadside of each intersection branch leading to the corresponding intersection, which is called the corresponding first roadside point. A first roadside point communication device 111 and multiple first monitoring devices 112 are pre-set on each first roadside point.

[0210] The remote visualization processing system for intersection data provided in Embodiment 2 of the present invention can execute the method steps in Embodiment 1 of the above method. Its implementation principle and technical effect are similar, and will not be described again here.

[0211] It should be noted that the above systems can be fully or partially integrated into a single physical entity, or they can be physically separated. They can be implemented entirely in software via processing element calls; they can be implemented entirely in hardware; or they can be partially implemented in software via processing element calls and partially in hardware. Furthermore, they can be stored in memory as program code, which can be called and executed by a processing element of the system. These functions can be integrated or implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. In the implementation process, the method steps of the aforementioned method or the processing steps of the aforementioned system can be completed through integrated logic circuits in the hardware of the processor element or through software instructions.

[0212] For example, the above system can be one or more integrated circuits configured to implement the aforementioned methods, such as one or more application-specific integrated circuits (ASICs), one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs). Furthermore, when a module of the above system is implemented through processing element scheduler code, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor capable of calling program code. Moreover, these modules can be integrated together to implement a system-on-a-chip (SOC).

[0213] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. This computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the foregoing method embodiments are generated. The computer described above can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The aforementioned computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the aforementioned computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, Bluetooth, microwave, etc.) means. The aforementioned computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The aforementioned available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state disks (SSDs)).

[0214] This invention provides a remote visualization processing method and system for intersection data. Multiple roadside points are set up at each intersection, and a set of monitoring devices and a roadside point communication device are installed at each roadside point. Each monitoring device connects to a remote cloud platform through its corresponding roadside point communication device. Each monitoring device monitors the traffic conditions of the current intersection and the current roadside point in real time and sends the monitoring data to the cloud platform. The cloud platform receives and stores the real-time monitoring data from the front end using an intersection monitoring database. Simultaneously, based on an asynchronous processing mechanism, it extracts information from the intersection monitoring database to perform 3D scene construction, traffic event analysis, traffic indicator analysis, and traffic participation value analysis, storing the dynamic analysis results in the intersection analysis database. Furthermore, it uses customized first and third visualization pages to display the real-time monitoring video, traffic event analysis, traffic indicator analysis, and traffic participation value analysis of each intersection, and uses a customized second visualization page based on digital twin technology to display the scene of the current intersection from the perspective of the current roadside point. This invention enhances the intelligent analysis level of road intersections and improves the real-time analysis capability of visual monitoring.

[0215] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0216] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented in hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0217] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for remote visualization processing of intersection data, characterized in that, The method includes: The cloud platform monitors the online status of equipment at each intersection and updates the intersection equipment database based on the monitoring results; The system receives real-time monitoring data from each intersection and stores it in the intersection monitoring database. The three-dimensional scene of each intersection is simulated based on the intersection monitoring database, and the simulation results are stored in the corresponding three-dimensional scene data table in the intersection analysis database; The traffic events at each intersection are analyzed in real time based on the intersection monitoring database, and the analysis results are stored in the corresponding traffic event data table in the intersection analysis database. The traffic indicators of each intersection are analyzed in real time based on the intersection monitoring database, and the analysis results are stored in the corresponding traffic indicator data table in the intersection analysis database. The number of traffic participants at each intersection is analyzed in real time based on the intersection monitoring database, and the analysis results are stored in the corresponding traffic participant data table in the intersection analysis database. The real-time traffic conditions of each intersection are visualized based on the real-time updated intersection equipment database, intersection monitoring database, intersection analysis database, as well as the pre-set operating vehicle database and intersection-roadside point relationship database. The traffic incident data table includes multiple first incident records. Each first incident record includes an eighth roadside point number field, a first incident type field, a first incident location field, a first incident impact range field, a first incident time field, and a first incident evidence video field. The first incident type field is the processing result obtained by performing event detection and classification on the telephoto camera video using a preset event classification model. The event classification model includes at least a pedestrian violation event analysis model, a non-motorized vehicle violation event analysis model, a motorized vehicle road violation event analysis model, and a motorized vehicle driving violation event analysis model. The time range of the first incident evidence video field is from 15 seconds before the event time to 15 seconds after the event time. The cloud platform loads the first, second, and third visualization pages during the process of visualizing the real-time traffic conditions at each intersection, and refreshes the content of the first, second, and third visualization pages at a preset refresh frequency. The intersection-roadside point relationship database includes multiple first relationship records; each first relationship record includes a first intersection number field, a first intersection name field, a first intersection center point coordinate field, and a first roadside point set field; the first roadside point set field is used to store the corresponding first roadside point set; the first roadside point set includes multiple first roadside point records; each first roadside point record includes a first roadside point number field, a first roadside point orientation field, and a first roadside point coverage area field. The intersection monitoring database includes multiple intersection monitoring sub-databases, each of which corresponds one-to-one with the first intersection number; each of the intersection monitoring sub-databases includes a video data table, a point cloud data table, an operating vehicle data table, and a traffic light pole data table; The video data table includes multiple first video records; each first video record includes a third side point number field, a second device number field, a second device type field, a first timestamp field, and a first video field; the second device type field includes telephoto cameras and panoramic cameras; The point cloud data table includes multiple first point cloud records; each first point cloud record includes a fourth side point number field, a third device number field, a third device type field, a second timestamp field, and a first radar point cloud field; the third device type field includes lidar and millimeter-wave radar. The step of simulating the 3D scene of each intersection based on the intersection monitoring database and storing the simulation results in the corresponding 3D scene data table in the intersection analysis database specifically includes: When the cloud platform adds a first point cloud record with the third device type field being LiDAR to any of the point cloud data tables, it takes the currently added first point cloud record as the corresponding type of point cloud record, extracts the second timestamp field of the type of point cloud record as the corresponding current timestamp, takes the point cloud data table with the newly added record as the corresponding current point cloud data table, takes the intersection monitoring sub-database and the first intersection number corresponding to the current point cloud data table as the corresponding current intersection monitoring sub-database and current intersection number, takes the video data table of the current intersection monitoring sub-database as the corresponding current video data table, extracts the fourth roadside point number field of the type of point cloud record as the corresponding current roadside point number, and extracts the first intersection center point coordinate field of the first relationship record in the intersection-roadside point relationship database that matches the first intersection number field with the current intersection number as the corresponding current intersection center point coordinate. Then, it performs intersection area map extraction processing from a preset high-precision map with the current intersection center point coordinate as the center to obtain the corresponding current intersection map. The first point cloud record in the current point cloud data table whose fourth side point number field is the current roadside point number, whose third device type field is millimeter-wave radar, and whose time interval between the second timestamp field and the current timestamp is less than a set time threshold is extracted as the corresponding second type point cloud record; and the first video record in the current video data table whose third side point number field is the current roadside point number, whose second device type field is panoramic camera, and whose time interval between the first timestamp field and the current timestamp is less than a set time threshold is extracted as the corresponding current panoramic video record; If neither the type II point cloud record nor the current panoramic video record is empty, then the first radar point cloud field of the type I and type II point cloud records is extracted as the corresponding type I and type II point clouds; and the first video field of the current panoramic video record is extracted as the corresponding current panoramic video; each point of the type I point cloud corresponds to a world coordinate and a reflection intensity; each point of the type II point cloud corresponds to a world coordinate and a relative velocity. Points with non-zero relative velocity in the second type of point cloud are considered moving points. Points in the first type of point cloud corresponding to each moving point and points exceeding the current intersection map coordinate range are deleted. The first type of point cloud with all points deleted is used as the corresponding current point cloud. Single-frame images are extracted from the current panoramic video to obtain multiple first-frame images. The image clarity of each first-frame image is identified, and the image with the highest clarity is selected as the corresponding current scene image. A bird's-eye view of the intersection is constructed based on the current intersection map to obtain a corresponding bird's-eye view scene image of the current intersection; multiple 3D first target detection boxes are obtained by performing target detection on the current point cloud based on a point cloud target detection model; multiple first target mask images with depth features are obtained by performing semantic segmentation processing on the current scene image based on a visual image segmentation model with depth estimation; a corresponding first matching group is formed by the first target detection boxes and the first target mask images corresponding to the same target; and the height, appearance, and color of the corresponding target in the current intersection bird's-eye view scene image are reconstructed in 3D based on each of the first matching groups to obtain a corresponding 3D scene image of the current intersection; each pixel of the current intersection 3D scene image inherits the corresponding world coordinates from the current intersection map through the current intersection bird's-eye view scene image; The first scene record, composed of the current roadside point number, the current timestamp, and the current intersection 3D scene map as corresponding fields, is added to the 3D scene data table of the intersection analysis sub-database corresponding to the current intersection number; The step of performing real-time analysis of traffic events at each intersection based on the intersection monitoring database and storing the analysis results in the corresponding traffic event data table within the intersection analysis database specifically includes: When the cloud platform adds the first video record with the second device type field being a telephoto camera to any of the video data tables, it takes the currently added first video record as the corresponding current video record, and extracts the three-way side point number field, the second device number field, and the first video field of the current video record as the corresponding current roadside point number, current device number, and current video. It also takes the video data table with the newly added record as the corresponding current video data table, and takes the intersection monitoring sub-database and the first intersection number corresponding to the current video data table as the corresponding current intersection monitoring sub-database and current intersection number. Based on a preset event classification model, the current video is processed for event detection and classification to obtain the corresponding event type, event location, event range, and event time. Then, the first video fields of all first video records in the current video data table, where the third roadside point number field matches the current roadside point number, the second device number field matches the current device number, and the first timestamp field is within a specified time range before and after the event time, are extracted and stitched together in chronological order to obtain the corresponding event evidence video. The corresponding first event record, composed of the current roadside point number, event type, event location, event range, event time, and event evidence video as corresponding fields, is added to the traffic event data table of the intersection analysis sub-database corresponding to the current intersection number. The event classification model includes at least a pedestrian violation event analysis model, a non-motorized vehicle violation event analysis model, a motorized vehicle road violation event analysis model, and a motorized vehicle driving violation event analysis model. After loading the first visualization page, the cloud platform retrieves a specified number of event records that most recently match the current roadside point number from the traffic event data table corresponding to the current intersection number; and sets the display content of the event type, event location, event range, and event time fields of each record in the roadside point traffic event list of the first visualization page based on the retrieved event records; sets a default viewing mark in the evidence viewing field of each record in the roadside point traffic event list; and when any of the viewing marks is clicked, plays the video in the first event evidence viewing video field of the first event record corresponding to the current viewing mark through a pop-up window. After loading the second visualization page, the cloud platform obtains the latest panoramic video and 3D scene map of the intersection from the video data table and the 3D scene data table corresponding to the current intersection number. It then performs visualization object simulation on the roadside equipment poles and their corresponding monitoring coverage areas, as well as the monitoring equipment on the poles and their corresponding monitoring equipment information, in the current 3D scene map to obtain a baseline 3D scene. It then iterates through each frame of the current panoramic video. During this iteration, it performs target segmentation on the current frame panoramic image based on a visual image segmentation model with depth estimation, and locates the coordinates of each target based on the conversion relationship between image coordinates and world coordinates. In the baseline 3D scene, it performs visualization object simulation on each identified person, animal, motor vehicle, or non-motor vehicle target to obtain the corresponding single-frame 3D scene. Finally, in the current frame 3D scene, it visualizes the motor vehicles belonging to operating vehicles. Above the traffic lights, a visual speed marker object is set to display the latest vehicle speed. Above each traffic light target location, a visual signal light sign object is set to display the latest signal light color, signal light type, and remaining signal light duration. Above each traffic light sign object, a visual traffic information sign object is set to display the latest lane traffic flow direction, lane flow rate, and lane queue length. The visualization information of each visual speed marker object and traffic light sign object is set based on the intersection monitoring sub-database corresponding to the current intersection, and the visualization information of each visual traffic information sign object is set based on the intersection analysis sub-database corresponding to the current intersection. After traversal, the scene sequence composed of all single-frame 3D scenes sorted chronologically is converted into a digital twin video, and the converted video is loaded and played on the second visualization page.

2. The remote visualization processing method for intersection data according to claim 1, characterized in that, Each intersection is assigned a unique intersection number, designated as the first intersection number. Each intersection includes multiple branch intersections, each branch intersection having a corresponding direction, including east, south, west, and north. If the number of branch intersections with a single direction is not unique, they are sequentially encoded based on the current direction. Motor vehicles traveling on each road at each intersection include both commercial and non-commercial vehicles. Each commercial vehicle is equipped with an OBU device. At each intersection, a roadside point is pre-set on the roadside leading to the corresponding intersection, denoted as the corresponding first roadside point; each first roadside point is assigned a unique roadside point number, denoted as the corresponding first roadside point number; each first roadside point is pre-set with a first roadside point communication device and multiple first monitoring devices. The first roadside communication device locally stores the corresponding first intersection number, first roadside number, and first roadside orientation; the first roadside orientation is consistent with the corresponding branch orientation. Each of the first monitoring devices is connected to the cloud platform through a corresponding first roadside point communication device; the first monitoring device locally stores a set of corresponding device parameters, including the first device name, the first device number, the first device type, and the first device manufacturer; the first device type includes telephoto cameras, panoramic cameras, lidar, millimeter-wave radar, RSU devices, and traffic signal poles; The first monitoring device, which is a telephoto camera, is used to capture real-time video of the monitored road to generate first real-time monitoring data containing a fixed-length video. The latest first real-time monitoring data is periodically sent to the cloud platform through the corresponding first roadside communication device at a preset synchronization frequency. The first real-time monitoring data includes a first timestamp, a first device number, a first data type, and first video data. The first data type is set to telephoto video type. The first monitoring device, which is a panoramic camera, is used to capture real-time video of the monitored road to generate a second real-time monitoring data containing a fixed-length video. The latest second real-time monitoring data is periodically sent to the cloud platform through the corresponding first roadside communication device at a preset synchronization frequency. The second real-time monitoring data includes a second timestamp, the first device number, a second data type, and second video data. The second data type is set to panoramic video type. The first monitoring device, which is a lidar device, is used to perform radar scanning on the monitoring environment to generate third real-time monitoring data. It periodically sends the latest third real-time monitoring data to the cloud platform via the corresponding first roadside communication device at a preset synchronization frequency. The third real-time monitoring data includes a third timestamp, the first device number, a third data type, and a first radar point cloud. The third data type is set to lidar point cloud type. The features of each point in the first radar point cloud include first coordinate features and first reflection intensity features. The coordinate system of the first coordinate features is the world coordinate system. The first monitoring device, which is a millimeter-wave radar, is used to perform radar scanning on the monitoring environment to generate a fourth real-time monitoring data, and periodically sends the latest fourth real-time monitoring data to the cloud platform through the corresponding first roadside point communication device at a preset synchronization frequency; the fourth real-time monitoring data includes a fourth timestamp, the first device number, a fourth data type, and a second radar point cloud; the fourth data type is set to millimeter-wave radar point cloud type; the features of each point in the second radar point cloud include a second coordinate feature and a first velocity feature; the coordinate system of the second coordinate feature is the world coordinate system; The first monitoring device, whose first device type is an RSU device, is used to receive first operating vehicle data sent by the OBU devices of each operating vehicle within the monitoring range, and to assemble all the first operating vehicle data received in the most recent first time period into corresponding fifth real-time monitoring data according to a preset first time period length; and to periodically send the latest fifth real-time monitoring data to the cloud platform through the corresponding first roadside point communication device at a preset synchronization frequency; the first operating vehicle data includes a first vehicle timestamp, a first vehicle license plate, a first vehicle model, a first vehicle color, a first driving mode, a first driver identifier, a first operating organization identifier, a first vehicle location, and a first vehicle speed; the first driving mode includes driverless driving, autonomous driving, and manual driving; the first driver identifier is the identity identifier of the current driver when the first driving mode is autonomous driving or manual driving; The fifth real-time monitoring data includes a fifth timestamp, the first device number, a fifth data type, and all first operating vehicle data received within the most recent first time period. The time interval between the first vehicle timestamps of any two first operating vehicle data within the fifth real-time monitoring data does not exceed the length of the first time period. The fifth data type is set as the operating vehicle type. The first monitoring device, whose first device type is a traffic signal pole, is used to acquire the real-time light status of all traffic lights on the pole, generate corresponding sixth real-time monitoring data, and periodically send the latest sixth real-time monitoring data to the cloud platform through the corresponding first roadside communication device at a preset synchronization frequency. The sixth real-time monitoring data includes a sixth timestamp, the first device number, a sixth data type, and multiple first traffic light data. The first traffic light data includes the first traffic light type, the first traffic light status, and the remaining duration of the first traffic light. The first traffic light type includes left turn light type, straight light type, and right turn light type. The first traffic light status includes red light status, yellow light status, yellow light flashing status, and green light status. The sixth data type is set to the light pole type. The first roadside point communication device is used to, when receiving real-time monitoring data sent by any type of the first monitoring device, take the first, second, third, fourth, fifth or sixth real-time monitoring data received at that time as the corresponding current real-time monitoring data, and send the corresponding first roadside point data packet composed of the first intersection number, the first roadside point number, the first roadside point orientation and the current real-time monitoring data to the cloud platform; The first roadside point communication device is also used to periodically detect whether the online status of all the first monitoring devices connected to it is normal, obtain the corresponding first device online status list, and send the first device heartbeat command carrying the first intersection number, the first roadside point number, and the first device online status list to the cloud platform; the first device online status list includes multiple first device status records; the first device status record includes a first monitoring device number field and a first monitoring device online status field; the first monitoring device online status field includes online status and offline status.

3. The remote visualization processing method for intersection data according to claim 2, characterized in that, The cloud platform includes the operating vehicle database, the intersection-roadside point relationship database, the intersection equipment database, the intersection monitoring database, and the intersection analysis database; The operating vehicle database includes multiple first vehicle records; each first vehicle record includes a first vehicle identifier field, a first vehicle license plate field, a first vehicle model field, a first vehicle color field, a first driving mode field, a first driver field, and a first operating organization field; the first driving mode field includes driverless, autonomous, and manual driving; the first driver field is empty when the first driving mode field is driverless, and is a designated driver identifier when the first driving mode field is autonomous or manual driving. The intersection equipment database includes multiple intersection equipment data tables, and each intersection equipment data table corresponds one-to-one with the first intersection number; The intersection equipment data table includes multiple first equipment records; each first equipment record includes a second roadside point number field, a first equipment number field, a first equipment name field, a first equipment type field, a first equipment manufacturer field, a first equipment status field, and a first equipment image field; the first equipment type field includes telephoto cameras, panoramic cameras, LiDAR, millimeter-wave radar, RSU equipment, and traffic signal poles; the first equipment status field includes online status and offline status. The operating vehicle data table includes multiple second vehicle records; each second vehicle record includes a fifth roadside point number field, a fourth device number field, a third timestamp field, a second vehicle license plate field, a second vehicle model field, a second vehicle color field, a second driving mode field, a second driver field, a second operating organization field, a first vehicle location field, and a first vehicle speed field; the second driving mode field includes driverless, autonomous, and manual driving; the second driver field is empty when the second driving mode field is driverless, and is the current driver's identity identifier when the second driving mode field is autonomous or manual driving; The traffic light pole data table includes multiple first pole records; each first pole record includes a sixth side point number field, a fifth device number field, a fourth timestamp field, and a first traffic light set field; the first traffic light set field is used to store the corresponding first traffic light set; the first traffic light set includes multiple first traffic light records; each first traffic light record includes a first traffic light type field, a first traffic light status field, and a first traffic light remaining duration field; the first traffic light type field includes left turn light, straight light, and right turn light; the first traffic light status field includes red light status, yellow light status, yellow light flashing status, and green light status. The intersection analysis database includes multiple intersection analysis sub-databases, each corresponding one-to-one with the first intersection number; each intersection analysis sub-database includes the 3D scene data table, the traffic event data table, the traffic indicator data table, and the traffic participant data table; The three-dimensional scene data table includes multiple first scene records; the first scene record includes a seventh side point number field, a fifth timestamp field, and a first intersection three-dimensional scene map field; The traffic indicator data table includes multiple first indicator records; the first indicator record includes the ninth roadside point number field, the fifth timestamp field, the first lane identifier field, the first lane traffic efficiency field, the first lane average speed field, the first lane average delay time field, the first lane average number of stops field, and the first lane average queue length field. The traffic participant data table includes multiple first participant records; each first participant record includes a tenth roadside point number field, a sixth timestamp field, a first type of traffic participant quantity field, a second type of traffic participant quantity field, and a third type of traffic participant quantity field; the traffic participant types corresponding to the first, second, and third type of traffic participant quantity fields are people, motor vehicles, and non-motor vehicles, respectively.

4. The remote visualization processing method for intersection data according to claim 3, characterized in that, The cloud platform monitors the online status of equipment at each intersection and updates the intersection equipment database based on the monitoring results, specifically including: The cloud platform assigns a corresponding first timer to each of the first roadside communication devices and times them at a normal time frequency. Upon receiving a heartbeat command from a first roadside communication device, the system resets the corresponding first timer and restarts the timing. It extracts the first intersection number, the first roadside point number, and the first device online status list from the heartbeat command as the corresponding current intersection number, current roadside point number, and current device online status list. It also uses the intersection device data table corresponding to the current intersection number as the corresponding current intersection device data table. Furthermore, it iterates through each first device status record in the current device online status list, extracting the first monitoring device number field and the first monitoring device online status field from the currently iterated first device status record as the corresponding current monitoring device number and current monitoring device online status. Finally, it resets the first device status field of the first device record in the current intersection device data table where the second roadside point number field matches the current roadside point number and the first device number field matches the current monitoring device number to the current monitoring device online status. The system will identify in real time whether the current timing result of each of the first timers exceeds the preset timing threshold. If it does, the system will take the first roadside point number and the first intersection number corresponding to the first roadside point communication device corresponding to the current first timer as the corresponding current offline roadside point number and current offline intersection number, and reset the first device status field of all the first device records in the intersection device data table corresponding to the current offline intersection number to the offline status.

5. The remote visualization processing method for intersection data according to claim 3, characterized in that, The process of receiving and storing real-time monitoring data from each intersection into the intersection monitoring database specifically includes: When the cloud platform receives the first roadside point data packet, it extracts the first intersection number, the first roadside point number, the first roadside point orientation, and real-time monitoring data as the corresponding current intersection number, current roadside point number, current roadside point orientation, and current real-time monitoring data. It then uses the video data table, point cloud data table, operating vehicle data table, and traffic light pole data table from the intersection monitoring sub-database corresponding to the current intersection number as the corresponding current video data table, current point cloud data table, current operating vehicle data table, and current traffic light pole data table. Furthermore, it extracts the timestamp and device number from the current real-time monitoring data as the corresponding current timestamp and current device number. Finally, it queries the intersection device data table corresponding to the current intersection number, extracts the first device type field from the first device record where the second roadside point number field matches the current roadside point number and the first device number field matches the current device number, and extracts the data type from the current real-time monitoring data as the corresponding current data type. When the current data type is telephoto video, the corresponding first video data is extracted from the current real-time monitoring data; and the corresponding first video record is added to the current video data table, consisting of the current roadside point number, the current device number, the current device type, the current timestamp, and the first video data as corresponding fields. When the current data type is panoramic video, the corresponding second video data is extracted from the current real-time monitoring data; and the corresponding first video record is added to the current video data table, consisting of the current roadside point number, the current device number, the current device type, the current timestamp, and the second video data as corresponding fields. When the current data type is a lidar point cloud, the corresponding first lidar point cloud is extracted from the current real-time monitoring data; and the corresponding first point cloud record is added to the current point cloud data table, consisting of the current roadside point number, the current device number, the current device type, the current timestamp, and the first lidar point cloud as corresponding fields. When the current data type is millimeter-wave radar point cloud, the corresponding second radar point cloud is extracted from the current real-time monitoring data; and the corresponding first point cloud record is added to the current point cloud data table, consisting of the current roadside point number, the current device number, the current device type, the current timestamp, and the second radar point cloud as corresponding fields. When the current data type is an operating vehicle type, multiple first operating vehicle data are extracted from the current real-time monitoring data; and the corresponding second vehicle record is added to the current operating vehicle data table, using the current roadside point number, the current device number, the current timestamp, and the first vehicle license plate, first vehicle model, first vehicle color, first driving mode, first driver identifier, first operating agency identifier, first vehicle location, and first vehicle speed of each first operating vehicle data as corresponding fields; When the current data type is a light pole type, multiple first traffic light data are extracted from the current real-time monitoring data; and the first traffic light type, first traffic light status, and remaining duration of each first traffic light data are used as corresponding fields to form a corresponding first traffic light record, and all the obtained first traffic light records are used to form a corresponding first traffic light set; and the current roadside point number, the current device number, the current timestamp, and the first traffic light set are used as corresponding fields to form a corresponding first light pole record, which is added to the current traffic light pole data table.

6. The remote visualization processing method for intersection data according to claim 3, characterized in that, The step of performing real-time analysis of traffic indicators at each intersection based on the intersection monitoring database and storing the analysis results in the corresponding traffic indicator data table within the intersection analysis database specifically includes: The cloud platform iterates through each of the intersection monitoring sub-databases at a specified time interval. During the iteration, the video data table and traffic light pole data table of the currently iterated intersection monitoring sub-database are used as the corresponding current video data table and current traffic light pole data table. The first intersection number corresponding to the currently iterated intersection monitoring sub-database is used as the corresponding current intersection number, and the traffic indicator data table of the intersection analysis sub-database corresponding to the current intersection number is used as the corresponding current traffic indicator data table. All first video records in the current video data table whose first timestamp field is within the most recent first specified time period and whose second device type field is a telephoto camera are extracted to form a corresponding set. The first video records in the set are then clustered by roadside point to obtain multiple first record sets. Similarly, all first light pole records in the current traffic light pole data table whose fourth timestamp field is within the most recent first specified time period are extracted to form a corresponding set. The first light pole records in the set are then clustered by roadside point to obtain multiple second record sets. A corresponding first set group is formed by the first and second record sets corresponding to the same roadside point. Based on the intersection number and roadside point number corresponding to the first set group, a high-precision road map of the corresponding road at the corresponding roadside point location under the corresponding intersection is obtained from a preset high-precision map as the corresponding first road map. All first video records in the first record set have the same third roadside point number field, all first light pole records in the second record set have the same sixth roadside point number field, and all records in the first set group have the same roadside point number field. The videos of the first video field of all the first video records in the first record set of the first set group are extracted and spliced ​​together in chronological order to obtain the corresponding first long video; the first long video is processed by single-frame image extraction to obtain multiple second frame images, and a target classification model based on visual images is used to perform vehicle target detection and classification processing on each second frame image to obtain multiple first vehicle target detection boxes to form a corresponding first frame vehicle target set; and a conventional target tracking algorithm is used to perform vehicle target tracking processing on all the first frame vehicle target sets to obtain multiple first vehicle trajectories; and the average speed of the corresponding first vehicle is obtained by dividing the trajectory length of each first vehicle trajectory by the trajectory duration; the first vehicle target detection box includes the detection box target type, detection box center coordinates, detection box size and detection box orientation angle, and the detection box target type is the vehicle type, which includes cars, buses, engineering vehicles, trucks and vans; Based on the second record set of the first set group, the time points when various traffic light types switched from other states to green light states within the most recent first specified time period are identified to obtain one or more first time points to form a corresponding first time point sequence; The first lane traffic index data set is obtained by estimating the average vehicle speed, traffic efficiency, and average delay time of each lane in the first road map. Specifically, the number of trajectory points of each first vehicle trajectory in each lane is counted to obtain the corresponding first lane point number, and the current first vehicle trajectory is taken as the subordinate trajectory of the lane with the largest number of first lane points; the average speed of the first vehicles in all first vehicle trajectories under each lane is averaged to obtain the corresponding first lane average speed; the traffic efficiency of each lane is calculated to obtain the corresponding first lane traffic efficiency = (first lane average speed / free flow speed) * 100%; and the average delay time of each lane is calculated to obtain the corresponding first lane average delay time = (intersection average length / first lane average speed) - (intersection average length / free flow speed); and the first lane traffic index data set is composed of the lane markings of each lane and the corresponding first lane average speed, first lane traffic efficiency, and first lane average delay time; the free flow speed is a preset fixed speed value, and the intersection average length is a preset fixed length value. The average number of stops and average queue length of each lane in the first road map are estimated to obtain the corresponding second lane traffic index data set. Specifically, a fixed vehicle length is assigned to each type of vehicle as the corresponding type vehicle length; each lane is traversed; during traversal, the currently traversed lane is recorded as the corresponding current lane, and the corresponding traffic light type of the current lane is obtained from the first road map and recorded as the current lane traffic light type. Each of the first time points in the first time point sequence corresponding to the current lane traffic light type is recorded as the corresponding second time point, and the number of the second time points is counted to generate the corresponding first total. Two counters starting with 0 and one data sequence starting with empty are initialized for the current lane and recorded as the corresponding first vehicle counter, first stop count counter, and first vehicle length sequence. It is confirmed whether each first vehicle trajectory has intersected with the current lane once or more. If it is confirmed that the trajectory has intersected once or more, the first vehicle counter is incremented. The counter value is incremented by 1; and it is confirmed whether each of the first vehicle trajectories is in the current lane at each of the second time points. If it is confirmed that the current first vehicle trajectory is in the current lane at the current second time point, the counter value of the first parking vehicle counter is incremented by 1, and the vehicle length corresponding to the vehicle type of the current first vehicle trajectory is added to the first vehicle length sequence; after the first vehicle counter and the first parking vehicle counter have completed counting and the first vehicle length sequence has also completed data addition, the corresponding first lane average number of stops is calculated as first parking vehicle counter / first vehicle counter, and the first vehicle length sequence is summed to generate the corresponding first vehicle length sum and the corresponding first lane average queue length is calculated as first vehicle length sum / first total; and at the end of the traversal, the lane identifier of each lane and the corresponding first lane average number of stops and first lane average queue length are combined to form the corresponding second lane traffic indicator data group; The current platform time is used as the current timestamp; and the corresponding first indicator record is added to the current traffic indicator data table by using the roadside point number corresponding to each of the first set groups, the current timestamp, the lane identifier corresponding to each lane, the first lane traffic efficiency, the first lane average speed, the first lane average delay time, the first lane average number of stops, and the first lane average queue length as corresponding fields.

7. The remote visualization processing method for intersection data according to claim 3, characterized in that, The step of performing real-time analysis of the number of traffic participants at each intersection based on the intersection monitoring database and storing the analysis results in the corresponding traffic participant data table within the intersection analysis database specifically includes: The cloud platform iterates through each of the intersection monitoring sub-databases at a specified time interval. During the iteration, the video data table of the currently iterated intersection monitoring sub-database is used as the corresponding current video data table, the first intersection number corresponding to the currently iterated intersection monitoring sub-database is used as the corresponding current intersection number, and the traffic participant data table of the intersection analysis sub-database corresponding to the current intersection number is used as the corresponding current traffic participant data table. Extract the first roadside point number fields from each of the first relationship records in the intersection-roadside point relationship database that match the first intersection number field with the current intersection number, and use them as the corresponding first numbers; extract the first timestamp field and first video field from the first video record in the current video data table that match the third roadside point number field with each of the first numbers, has the second device type field as a panoramic camera, and has the first timestamp field as the one closest to the current time, and use them as the corresponding first panoramic timestamp and second panoramic video; extract single-frame images from each of the second panoramic videos and use the last frame as the corresponding first panoramic image; and use a visual image-based target classification model to process the first panoramic image. Multiple second target detection boxes are obtained through target detection and classification. The total number of second target detection boxes for target types (people, motor vehicles, and non-motor vehicles) is counted to obtain the corresponding number of first, second, and third traffic participants. The first participant record is added to the current traffic participant data table by using the first number, the corresponding first panoramic timestamp, and the number of first, second, and third traffic participants as corresponding fields. The second target detection box includes target type, target coordinates, target size, and target orientation angle. Target types include people, animals, motor vehicles, non-motor vehicles, and buildings. The second target detection box includes the detection box target type, detection box center coordinates, detection box size, and detection box orientation angle.

8. The remote visualization processing method for intersection data according to claim 3, characterized in that, The method further includes: pre-setting three visualization pages, namely a first, second, and third visualization page; wherein... The first visualization page includes a first, second, third, and fourth display area. The first display area includes an intersection name entry and multiple intersection number entries. The second display area includes a total number of intersection monitoring devices and multiple entries for the number of first-class devices. The third display area includes a map area, a total number of roadside point monitoring devices, and a list of roadside point monitoring devices. Each record in the roadside point monitoring device list includes the device name, device number, device orientation, and device manufacturer fields. The fourth display area includes a list of roadside point traffic events. Each record in the roadside point traffic event list includes the event type, event location, event range, event time, and evidence viewing fields. The second visualization page is a visualization page implemented based on digital twin technology; The third visualization page includes a monitoring video area, a first analysis area, and a second analysis area. The monitoring video area includes east, south, west, and north video areas and an intersection twin video area. The intersection twin video area is a video area implemented based on digital twin technology, including orientation entries and comparison markers. The first analysis area includes traffic efficiency entries, average vehicle speed entries, average delay time entries, average number of stops entries, average queue length entries, and an envelope diagram display area. The second analysis area includes a statistical chart display area, pedestrian quantity entries, motor vehicle quantity entries, and non-motor vehicle quantity entries.

9. The remote visualization processing method for intersection data according to claim 8, characterized in that, The process of visualizing the real-time traffic conditions at each intersection based on the real-time updated intersection equipment database, intersection monitoring database, intersection analysis database, and pre-set operating vehicle database and intersection-roadside point relationship database specifically includes: The cloud platform loads the first visualization page; The total number of the first relationship records in the intersection-roadside point relationship database is counted to obtain the corresponding total number of first intersections; and an intersection number entry for the total number of first intersections is created in the first display area of ​​the first visualization page. All intersection number entries are browsed by flipping through the triangular page-turning symbols on the left and right sides of the first display area. A one-to-one correspondence is established between each intersection number entry and the first relationship record. The display content of the corresponding intersection number entry is set by the first intersection number field of each first relationship record. After the setting is completed, the first intersection number entry is selected as the currently selected intersection number entry. When any of the intersection number entries is selected, the first relationship record corresponding to the currently selected intersection number entry is taken as the corresponding current relationship record, and the first intersection number field of the current relationship record is extracted as the corresponding current intersection number. The display content of the intersection name entry in the first display area is set according to the first intersection name field of the current relationship record. In the intersection equipment data table corresponding to the current intersection number, the number of equipment types is statistically analyzed to generate a corresponding first type quantity. The total number of the first equipment records is statistically analyzed to obtain the corresponding first equipment total quantity. The total number of the first equipment records of each type of the same equipment is statistically analyzed to obtain the first type quantity of the first category of equipment. The display content of the intersection monitoring equipment total quantity entry in the second display area is set according to the first equipment total quantity. An entry for the first type of equipment quantity is created in the second display area, and a one-to-one correspondence is established between each entry for the first type of equipment quantity and the total number of the first type of equipment. The corresponding first type of equipment quantity information is composed of each total number of the first type of equipment and the corresponding equipment type name. The display content of the corresponding entry for the first type of equipment quantity is set according to the first type of equipment quantity information. Based on the center point coordinate field of the first intersection in the current relationship record, the corresponding high-precision map of the intersection is extracted from the preset high-precision map and loaded into the map area of ​​the third display area; based on the orientation field of the first roadside point records of the first roadside point set in the current relationship record, roadside point marking is drawn on the high-precision map of the intersection in the map area; when any roadside point mark is selected, the currently selected roadside point mark is enlarged to generate the corresponding current roadside point mark; a prompt box is used above the current roadside point mark to provide a prompt description of the current roadside point; and the first roadside point number field corresponding to the current roadside point mark in the current relationship record is extracted as the corresponding current roadside point number; In the intersection equipment data table corresponding to the current intersection number, all first equipment records that match the second roadside point number field with the current roadside point number are extracted to form a corresponding first record list; the total number of first equipment records in the first record list is counted to obtain the corresponding second total number of equipment; the display content of the total number of roadside point monitoring equipment entries in the third display area is set based on the second total number of equipment; and the content of each record in the roadside point monitoring equipment list in the third display area is set based on the first record list. In the traffic event data table of the intersection analysis sub-database corresponding to the current intersection number, all the first event records that most recently match the ninth roadside point number field with the current roadside point number are extracted to form a corresponding second record list; and based on the first event type field, first event location field, first event impact range field, and first event time field of each record in the second record list, the display content of the event type, event location, event range, and event time fields of each record in the roadside point traffic event list in the fourth display area is set; and a default viewing mark is set in the evidence viewing field of each record in the roadside point traffic event list; and when any of the viewing marks is clicked, the video in the first event evidence viewing video field of the first event record corresponding to the current viewing mark in the second record list is played through a pop-up window; The cloud platform loads the second visualization page; The first video field, which matches the third roadside point number field with the current roadside point number, has the second device type field being the panoramic camera, and the first timestamp field being the first video record closest to the current time, is extracted from the video data table of the intersection monitoring sub-database corresponding to the current intersection number as the corresponding current panoramic video. The first intersection 3D scene map field, which matches the seventh roadside point number field with the current roadside point number, and has the fifth timestamp field being the first scene record closest to the current time, is extracted from the 3D scene data table of the intersection analysis sub-database corresponding to the current intersection number as the corresponding first intersection 3D scene map. The system simulates corresponding roadside equipment poles at various roadside points in the 3D scene of the first intersection using preset roadside equipment pole visualization objects; it also simulates monitoring equipment on each roadside equipment pole visualization object using preset monitoring equipment visualization objects, and sets the icons and names of each monitoring equipment visualization object according to the monitoring equipment information provided by the intersection equipment data table corresponding to the current intersection number; furthermore, it simulates the monitoring coverage area under each roadside equipment pole visualization object using preset roadside point coverage area visualization objects, and sets the size of each roadside point coverage area visualization object according to the coverage area field of each first roadside point in the current relationship record; and uses the resulting 3D scene of the first intersection as the corresponding baseline 3D scene. The current panoramic video is processed to extract multiple first-frame panoramic images by extracting single-frame images. Each first-frame panoramic image is then traversed, and during traversal, the currently traversed first-frame panoramic image is used as the corresponding current panoramic image. Semantic segmentation processing is performed on the current panoramic image based on a visual image segmentation model with depth estimation to obtain multiple second target mask images with depth features. The world coordinates of each second target mask image are obtained based on the transformation relationship between image coordinates and world coordinates. In the baseline 3D scene, at the scene location corresponding to the world coordinates of the second target mask image representing each target semantic type (human, animal, motor vehicle, or non-motorized vehicle), a corresponding human, animal, motorized vehicle, or non-motorized vehicle is created. The system visualizes vehicles or non-motorized vehicles and sets their appearance based on the image features of the second target mask image. The set baseline 3D scene is recorded as the corresponding first frame 3D scene. Based on the first vehicle positioning field of all second vehicle records in the operating vehicle data table of the intersection monitoring sub-database corresponding to the current intersection number that are closest to the current panoramic image and have a time interval less than a set threshold, the system identifies whether each visualized motor vehicle in the first frame 3D scene is an operating vehicle. If so, the first vehicle speed field of the corresponding second vehicle record is extracted as the corresponding first target speed, and a preset visualization speed marker is used. The object simulates speed prompt information above the visualized motor vehicle object, and sets the speed prompt information of the visualized speed marker object based on the first target vehicle speed; and creates a traffic light sign visualized object above the scene position corresponding to the world coordinates of the second target mask image with the semantic type of traffic light in the first frame 3D scene, and sets the display content of the traffic light color, traffic light type and remaining traffic light duration of each corresponding traffic light sign visualized object based on the first light pole record at each roadside point in the traffic light pole data table of the intersection monitoring sub-database corresponding to the current intersection number, which is closest to the time of the current panoramic image; and in the In the first frame of the 3D scene, a traffic information sign visualization object is created above each of the traffic light sign visualization objects. Based on the intersection high-precision map and the traffic indicator data table of the intersection analysis sub-database corresponding to the current intersection number, the first indicator record at each roadside point that is closest to the time of the current panoramic image is used to set the display content of lane traffic direction, lane flow rate, and lane queue length for each of the traffic information sign visualization objects. At the end of the traversal, all the obtained first frame 3D scenes are sorted in chronological order to obtain the corresponding first frame 3D scene sequence. Digital twin video conversion is performed on the first frame 3D scene sequence to obtain the corresponding first twin video.Each of the second target mask images corresponds to a target semantic type, and the target semantic type includes people, animals, motor vehicles, non-motor vehicles, and traffic lights; The second visualization page loads and plays the first twin video; if the user selects any of the monitoring device visualization objects during playback, a device sign visualization object is created on the selected monitoring device visualization object, and the device sign visualization object displays the name, online status, device image, type, number, orientation, and manufacturer of the current monitoring device according to the current relationship record and the intersection device data table corresponding to the current intersection number; The cloud platform loads the third visualization page; Then, from the video data table of the intersection monitoring sub-database corresponding to the current intersection number, the first video field of the first video record with the second device number field being a telephoto camera, which is the latest in time and the time interval between each roadside point does not exceed a preset time threshold, is extracted as the corresponding first roadside point real-time video; and according to the roadside point orientation corresponding to each first roadside point real-time video, the corresponding video areas in the four video areas of east, south, west and north within the monitoring video area are loaded and played; and after any of the four video areas is selected, the corresponding video area is loaded and played. The orientation is set as the current orientation, and the real-time video of the first roadside point currently playing in the current video area is set as the corresponding real-time video of the current roadside point. The current roadside point number is modified to match the roadside point number corresponding to the current video area. In the intersection twin video area of ​​the monitoring video area, the real-time video of the current roadside point is rendered using digital twin technology. The orientation entry in the intersection twin video area is set to the corresponding current orientation. When the comparison mark in the intersection twin video area is clicked, the real-time video of the current roadside point and the corresponding rendered video are switched and compared in the intersection twin video area. Then, from the traffic indicator data table of the intersection analysis sub-database corresponding to the current intersection number, the first lane traffic efficiency field, the first lane average speed field, the first lane average delay time field, the first lane average number of stops field, and the first lane average queue length field of each lane corresponding to the current roadside point number are extracted as the corresponding first traffic efficiency, first average speed, first average delay time, first average number of stops field, and first average queue length; and the first traffic efficiency, first average speed, first average delay time, and first average number of stops field for all lanes are calculated. The average values ​​of the first roadside point traffic efficiency, the first roadside point average vehicle speed, the first roadside point average delay time, the first roadside point average number of stops, and the first roadside point average queue length are calculated by averaging the first average queue length. Based on the first roadside point traffic efficiency, the first roadside point average vehicle speed, the first roadside point average delay time, the first roadside point average number of stops, and the first roadside point average queue length, the display content of the traffic efficiency item, the average vehicle speed item, the average delay time item, the average number of stops item, and the average queue length item in the first analysis area are set. When the traffic efficiency item is selected, the traffic efficiency of the first roadside point at each time point within the most recent first specified time period is calculated based on the historical data of the current traffic indicator data table to obtain the corresponding first roadside point traffic efficiency sequence. A first envelope diagram is then plotted in the envelope diagram display area with time as the horizontal axis and efficiency percentage as the vertical axis based on the first roadside point traffic efficiency sequence. When the average vehicle speed item is selected, the average vehicle speed of the first roadside point at each time point within the most recent first specified time period is calculated based on the historical data of the current traffic indicator data table to obtain the corresponding first roadside point average vehicle speed sequence. A second envelope diagram is then plotted in the envelope diagram display area with time as the horizontal axis and vehicle speed as the vertical axis based on the first roadside point average vehicle speed sequence. When the average delay time item is selected, the average delay time of the first roadside point at each time point within the most recent first specified time period is calculated based on the historical data of the current traffic indicator data table to obtain the corresponding first roadside point average delay time. The system calculates the average delay time series of the first roadside points and plots a third envelope diagram in the envelope diagram display area with time as the horizontal axis and delay time as the vertical axis. When the average number of stops is selected, it calculates the average number of stops at the first roadside points at each time point within the most recent first specified time period based on the historical data of the current traffic indicator data table. Based on this first roadside point average number of stops sequence, it plots a fourth envelope diagram in the envelope diagram display area with time as the horizontal axis and number of stops as the vertical axis. When the average queue length is selected, it calculates the average queue length at the first roadside points at each time point within the most recent first specified time period based on the historical data of the current traffic indicator data table. Based on this first roadside point average queue length sequence, it plots a fifth envelope diagram in the envelope diagram display area with time as the horizontal axis and queue length as the vertical axis. Then, from the traffic participant data table of the intersection analysis sub-database corresponding to the current intersection number, the number fields of the first, second, and third types of traffic participants that match the current intersection number in the tenth roadside point number field and whose sixth timestamp field is the most recent time are extracted as the corresponding first pedestrian number, first motor vehicle number, and first non-motor vehicle number; and the display content of the pedestrian number entry, motor vehicle number entry, and non-motor vehicle number entry in the second analysis area is set based on the first pedestrian number, first motor vehicle number, and first non-motor vehicle number. When the pedestrian number item, the motor vehicle number item, or the non-motor vehicle number item is selected, the number of the first pedestrian, the number of the first motor vehicle, or the number of the first non-motor vehicle at each time point within the most recent second specified time period is obtained based on the historical data of the current traffic participant data table, thereby forming the corresponding first pedestrian number sequence, first motor vehicle number sequence, or first non-motor vehicle number sequence. Based on the first pedestrian number sequence, the first motor vehicle number sequence, or the first non-motor vehicle number sequence, the corresponding first, second, or third curve is plotted in the statistical chart display area with time as the horizontal axis and quantity as the vertical axis. The cloud platform refreshes the content of the first, second, and third visualization pages at a preset refresh frequency.

10. A system for implementing the remote visualization processing method for intersection data according to any one of claims 1-9, characterized in that, The system includes: a cloud platform, multiple first-path side-point communication devices, and multiple first-path monitoring devices; each of the first-path monitoring devices is connected to the cloud platform through a corresponding first-path side-point communication device. The cloud platform is used to monitor the online status of equipment at each intersection and update the intersection equipment database based on the monitoring results; receive real-time monitoring data from each intersection and store it in the intersection monitoring database; simulate the 3D scene of each intersection based on the intersection monitoring database and store the simulation results in the corresponding 3D scene data table in the intersection analysis database; perform real-time analysis of traffic events at each intersection based on the intersection monitoring database and store the analysis results in the corresponding traffic event data table in the intersection analysis database; perform real-time analysis of traffic indicators at each intersection based on the intersection monitoring database and store the analysis results in the corresponding traffic indicator data table in the intersection analysis database; perform real-time analysis of the number of traffic participants at each intersection based on the intersection monitoring database and store the analysis results in the corresponding traffic participant data table in the intersection analysis database; and visualize the real-time traffic conditions of each intersection based on the real-time updated intersection equipment database, intersection monitoring database, intersection analysis database, and the pre-set operating vehicle database and intersection-roadside point relationship database.

11. The system according to claim 10, characterized in that, Each intersection includes multiple intersection branches. A roadside point is pre-set on the roadside of each intersection branch leading to the corresponding intersection, and is denoted as the corresponding first roadside point. A first roadside point communication device and multiple first monitoring devices are pre-set on each first roadside point.

Images