A roadside parking monitoring method and system based on high-bit video

Abstract

The application provides a roadside parking monitoring method and system based on high-position video, relates to the technical field of intelligent monitoring, and constructs a parking tendency monitoring area and a roadside parking monitoring area, and respectively arranges a first high-position video collection gun and a second high-position video collection gun; the first high-position video collection gun performs vehicle driving dynamic analysis on vehicle flow, determines a target monitoring vehicle and information, activates the second high-position video collection gun based on the target monitoring vehicle; the second high-position video collection gun performs parking time identification and vehicle driving-out analysis on the target monitoring vehicle, generates a driving-out time identification; a parking charging analysis module generates target vehicle charging information; target vehicle charging information is sent according to a payment contact channel obtained according to vehicle owner information. The technical problems of inaccurate parking time and low reliability of parking charging results in the prior art are solved, accurate control of parking time is realized, the reliability of parking charging results is improved, and the technical effect of intelligent parking monitoring is achieved.

Description

Technical Field

[0001] This invention relates to the field of monitoring technology, specifically to a roadside parking monitoring method and system based on high-position video. Background Technology

[0002] The supply and demand imbalance of on-street parking spaces remains prominent, and the problem of parking difficulties is intensifying. Currently, most roadside parking areas still use a combination of methods such as video parking stations, manual payment, and geomagnetic sensors + PDAs. Parking fees are often determined based on the analysis of large amounts of vehicle monitoring data, which leads to technical problems such as inaccurate parking time and low reliability of parking fee results. High-position video parking systems, on the other hand, use video analysis algorithms to detect whether surrounding parking spaces are occupied. By setting up high-position video capture guns to analyze the driving dynamics of target vehicles, they achieve precise control of parking time and improve the reliability of parking fee results. Summary of the Invention

[0003] This application provides a roadside parking monitoring method and system based on high-position video, which addresses the technical problems of inaccurate parking time and low reliability of parking billing results in the prior art.

[0004] In view of the above problems, this application provides a roadside parking monitoring method and system based on high-position video. By setting up a high-position video acquisition gun to perform dynamic analysis of the target vehicle's driving, it achieves accurate control of parking time, improves the reliability of parking billing results, and achieves the technical effect of intelligent parking monitoring.

[0005] In a first aspect, embodiments of this application provide a roadside parking monitoring method based on high-position video. The method includes pre-constructing a parking tendency monitoring area and a roadside parking monitoring area. The parking tendency monitoring area is equipped with a first high-position video capture gun, and the roadside parking monitoring area is equipped with a second high-position video capture gun. Based on the first high-position video capture gun, the method performs vehicle dynamic analysis on the traffic flow entering the parking tendency monitoring area to determine target monitored vehicles and their information. The target vehicle information includes license plate number, brand and model, appearance characteristics, and owner information. The method activates the second high-position video capture gun based on the target monitored vehicle. The second high-position video capture gun identifies the parking time of the target monitored vehicle entering the roadside parking monitoring area. The method analyzes the vehicle's exit and generates an exit time identifier. A parking fee analysis module is pre-set, and the parking time identifier, exit time identifier, and target vehicle information are sent to the parking fee analysis module for fee statistics to generate target fee information. The method obtains a payment contact channel based on the owner information and sends the target fee information based on the payment contact channel.

[0006] Secondly, embodiments of this application provide a roadside parking monitoring system based on high-position video. The system includes: a pre-construction module for parking tendency monitoring areas and roadside parking monitoring areas, wherein the parking tendency monitoring areas are equipped with a first high-position video acquisition gun, and the roadside parking monitoring areas are equipped with a second high-position video acquisition gun; a vehicle driving dynamic analysis module, which performs vehicle driving dynamic analysis on the traffic flow entering the parking tendency monitoring areas based on the first high-position video acquisition gun, determining target monitored vehicles and target vehicle information, wherein the target vehicle information includes license plate number, brand and model, appearance characteristics, and owner information; and a second high-position video acquisition gun activation module, which is used for... The system activates the second high-position video capture gun based on the target monitored vehicle; a target monitored vehicle parking time identification module is used by the second high-position video capture gun to identify the parking time of the target monitored vehicle entering the roadside parking monitoring area; a vehicle exit analysis module is used to analyze the vehicle exit of the target monitored vehicle and generate an exit time identifier; a parking fee analysis module is used to preset the parking fee analysis module and send the parking time identifier, the exit time identifier, and the target vehicle information to the parking fee analysis module for fee statistics and to generate target fee information; and a fee information sending module is used to obtain a payment contact channel based on the vehicle owner information interaction and send the target fee information based on the payment contact channel.

[0007] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0008] This application embodiment constructs a parking tendency monitoring area and a roadside parking monitoring area, and deploys a first high-position video capture gun and a second high-position video capture gun respectively. The first high-position video capture gun is used to perform vehicle dynamic analysis on the traffic flow entering the parking tendency monitoring area to determine the target monitored vehicles and their information. Then, based on the determined target monitored vehicles, the second high-position video capture gun is activated. The second high-position video capture gun is used to mark the parking time of the target monitored vehicles entering the roadside parking monitoring area and analyze vehicle exits to generate exit time markers. Finally, the parking fee analysis module generates target parking fee information; based on the vehicle owner's information interaction to obtain the payment contact channel, the target parking fee information is sent. This invention solves the technical problems of inaccurate parking time and low reliability of parking fee results in the prior art, achieving precise control of parking time, improving the reliability of parking fee results, and achieving the technical effect of intelligent parking monitoring.

[0009] The above description is merely an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, specific embodiments of this application are given below. Attached Figure Description

[0010] Figure 1 This application provides a schematic flowchart of a roadside parking monitoring method based on high-position video.

[0011] Figure 2 This application provides a schematic diagram of the activation process of the second high-position video acquisition gun in a roadside parking monitoring method based on high-position video.

[0012] Figure 3 This application provides a structural block diagram of a roadside parking monitoring system based on high-position video.

[0013] The attached diagram shows the following modules: parking tendency monitoring area and roadside parking monitoring area pre-construction module, vehicle driving dynamic analysis module, second high-position video acquisition gun activation module, target monitoring vehicle parking time identification module, vehicle exit analysis module, parking fee analysis module, and parking fee information sending module. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the described embodiments are only a part of the embodiments of this application, and not all of them, and this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. Furthermore, it should be noted that, for ease of description, only the parts relevant to this application are shown in the accompanying drawings, not all of them.

[0015] Example 1

[0016] like Figure 1 As shown in the figure, this application provides a roadside parking monitoring method based on high-position video, the method including:

[0017] A parking tendency monitoring area and a roadside parking monitoring area are pre-constructed, wherein the parking tendency monitoring area is equipped with a first high-position video acquisition gun and the roadside parking monitoring area is equipped with a second high-position video acquisition gun;

[0018] Specifically, the parking tendency monitoring area refers to the area away from roadside parking spaces, while the roadside parking monitoring area refers to the roadside parking area that can be captured by high-position video capture guns. High-position video capture guns are intelligent image monitoring hardware devices installed high up within the road, such as streetlight poles or public security monitoring poles. These high-position video capture guns are used to determine the status of vehicles (entering and exiting) in the roadside parking monitoring area, capturing images of the front and rear of vehicles. They consist of two parts: the capture gun and the camera lens. The capture gun is a type of CCD camera used in surveillance, rectangular in shape, without a lens. It uses a rotating lens to capture images from different angles, zooming in and out for clearer images. The first high-position video capture gun is deployed in the parking tendency monitoring area to identify whether a vehicle has a parking tendency by judging the vehicle's movement trajectory within the monitoring area. The second high-position video capture gun in the roadside parking monitoring area is used to judge the status of vehicles (entering and exiting) in the roadside parking monitoring area and generate parking fee information.

[0019] Based on the first high-position video acquisition gun, the vehicle driving dynamics of the traffic entering the parking tendency monitoring area are analyzed to determine the target monitoring vehicle and the target vehicle information, wherein the target vehicle information includes license plate number, brand and model, appearance characteristics and owner information;

[0020] Specifically, vehicle driving dynamic analysis refers to analyzing the speed or trajectory of vehicles entering the parking tendency monitoring area using a first high-position acquisition camera. This includes analyzing whether there are changes in speed (deceleration) and vehicle tire trajectory (tilt). If a parking tendency is detected, the first high-position video acquisition camera will focus on monitoring the vehicle as a target vehicle. The target vehicle information includes license plate number, brand and model, appearance characteristics, and owner information. The license plate number refers to the license plate number captured in the photograph, the vehicle brand refers to the vehicle logo captured in the photograph, the appearance characteristics refer to the shape and outline of the vehicle, and the owner information is based on the registration information of the target vehicle obtained by the traffic management department.

[0021] The second high-position video acquisition gun is activated based on the target monitored vehicle;

[0022] Specifically, the first high-position acquisition gun monitors vehicles in the parking tendency monitoring area to identify target vehicles with parking tendency. Then, the target vehicles are monitored closely. If the target vehicle parks, the second high-position acquisition gun is activated to monitor the vehicle status (entering the roadside parking area). If the target vehicle does not park, the second high-position acquisition gun does not need to be activated.

[0023] The second high-position video capture camera marks the parking time of the target monitored vehicle that enters the roadside parking monitoring area;

[0024] Specifically, the second high-position video capture gun monitors target vehicles in the roadside parking area. If a target vehicle enters a parking space, the second high-position capture gun monitors and records the target vehicle's parking time, generating a parking time identifier.

[0025] Perform vehicle departure analysis on the target monitored vehicle and generate departure time identifier;

[0026] Specifically, if the target vehicle in the roadside parking area changes its status, such as starting up, the target vehicle's speed and trajectory will change. This will activate the second high-position acquisition gun to monitor and photograph the target vehicle and record its departure time, generating a departure time identifier.

[0027] A parking fee analysis module is preset, and the parking time identifier, the exit time identifier and the target vehicle information are sent to the parking fee analysis module to perform fee statistics and generate target fee information.

[0028] Specifically, the parking fee analysis module calculates the parking fee for the target vehicle based on the parking time and departure time indicators captured by the second high-position sensor, according to the parking fee rules. For example, the fee rules could be: 3 yuan for the first half hour, 6 yuan per half hour for the first three hours, and 10 yuan per half hour after three hours. If the parking time is half an hour, the parking fee analysis module calculates the parking fee as 3 yuan according to the rules, and then generates target parking fee information based on the target vehicle information captured by the first high-position sensor. This target vehicle information includes the vehicle's license plate (whether it's a counterfeit), the vehicle's brand and appearance, and the owner's information. The target parking fee information can include the location of the roadside parking area, the parking fee payable, and the parking time.

[0029] The payment contact channel is obtained through the interaction of the vehicle owner information, and the target fare information is sent based on the payment contact channel.

[0030] Specifically, the payment contact information of the vehicle owner is obtained based on the vehicle owner information registered with the vehicle management office. The payment contact information may include the vehicle owner's mobile phone number, Alipay, email, etc. Based on the contact information, the payment information can be sent to the vehicle owner via SMS, QR code, link, or email.

[0031] Furthermore, the steps in this application, based on the first high-position video acquisition gun performing vehicle dynamic analysis on the traffic flow entering the parking tendency monitoring area to determine the target monitored vehicle and target vehicle information, also include:

[0032] The parking tendency monitoring area is captured by the first high-position video acquisition gun to obtain the target traffic flow video.

[0033] The target traffic flow video is processed to extract video frames to obtain the first video frame;

[0034] The first frame of video is synchronized to the vehicle segmentation module to obtain K vehicles waiting to stop, wherein the K vehicles waiting to stop correspond to K vehicle code identifiers, and K is a positive integer;

[0035] Based on the target traffic video, multi-target tracking of the K vehicles waiting to stop is performed to obtain K features of the vehicles waiting to stop, wherein the features of the vehicles waiting to stop include speed features and trajectory features;

[0036] The features of the K vehicles waiting to be parked are synchronized to the parking tendency analysis subnetwork to obtain K parking tendency indices. The target monitored vehicle is obtained by serializing the K parking tendency indices and calling the vehicle waiting to be parked corresponding to the largest parking tendency index.

[0037] The key features of the target monitored vehicle are extracted to obtain the target vehicle information.

[0038] Specifically, the target traffic flow video refers to the video of traffic flow within a parking tendency monitoring area over a certain period of time, captured by the first high-position acquisition gun. This traffic flow video consistently contains target vehicles. Video frame extraction of the target traffic flow video involves capturing multiple frames from the video. The first frame is obtained by extracting multiple frames from the target traffic flow video. This first frame is the image among the multiple images captured from the target traffic flow video that clearly identifies the target vehicle information and the driver's profile. The first frame should have high clarity and be free of distortion. The vehicle segmentation module separates the target vehicle from the road background, leaving only the information of the vehicle waiting to stop. In this embodiment, the first frame should contain information about the vehicle waiting to stop and background information of non-target vehicles. The constructed vehicle segmentation module automates the separation of the image of the vehicle waiting to stop and the background image in the acquired first frame video, reducing interference from irrelevant backgrounds in the evaluation of the vehicle waiting to stop. The preferred vehicle segmentation module is as follows: Information about the vehicles to be parked is obtained. In this embodiment, K pieces of information about the vehicles to be parked are obtained based on big data or data from the vehicle management office and the vehicle manufacturer. Each piece of information includes its corresponding license plate information, vehicle exterior outline information, owner information, etc. Experienced personnel perform manual semantic segmentation on the first frame video image of the vehicles to be parked. Specifically, the images of the vehicles to be parked and the background image in the first frame video image are divided and segmented, and the segmentation results are labeled with "vehicle to be parked" and "background image" to obtain a first segmentation result set for the first frame video image. Each segmentation result of the vehicles to be parked in the first segmentation result set corresponds one-to-one with each first frame video image of the target vehicle in the first frame video image set. Each first segmentation result of the first frame video image of the vehicles to be parked is an image composed of two semantic region blocks with semantic labels "vehicle to be parked" and "background image". The preferred method for constructing the vehicle segmentation module is based on a fully convolutional neural network (WCNN). The advantage of WCNNs is that they do not strictly limit the size of the first input video frame. In the vehicle segmentation module, the convolutional and pooling layers in the encoder are used to progressively extract semantic features from the image (the first frame video image of the vehicle to be stopped) and progressively reduce the size of the feature map, ultimately compressing the high-resolution first grid image into a low-resolution feature map. The depooling and deconvolutional layers in the decoder are used to restore the size of the low-resolution feature map, while the decoder generates pixel-level class probability distributions, thereby achieving semantic segmentation of the first frame video image of the vehicle to be stopped. After synchronizing the first frame video of the target vehicle to the vehicle segmentation module for classification and segmentation, K vehicles to be stopped are obtained, and K target vehicles are encoded (1-K). Here, the encoding of the vehicles to be stopped is inconsistent with the license plate number of the vehicles to be stopped. After obtaining the target vehicle, the corresponding license plate number is obtained through encoding to reduce the waste of computing resources.Based on the target traffic flow video, K vehicles waiting to stop are simultaneously tracked to obtain K features of the vehicles waiting to stop, including the acceleration and trajectory of the vehicles. The parking tendency analysis subnetwork analyzes the characteristics of K vehicles waiting to park, such as acceleration, driving trajectory, and curvature, to determine whether the speed has slowed down and whether the vehicle's driving trajectory has changed. During the analysis, the speed analysis results and driving trajectory can be weighted. In this example, the speed analysis results and driving trajectory analysis results are preferably weighted at a ratio of 3:7. Based on the weighted results, K parking tendency indices are output. The vehicle with the highest parking tendency index is selected as the target monitoring vehicle. Then, the first high-resolution acquisition camera extracts key features of the target monitoring vehicle to obtain target vehicle information, including the vehicle's license plate number, appearance features, vehicle brand, and owner information. Appearance features can be identified through the vehicle's outline. A vehicle brand database is constructed by collecting logos from different brands, and the vehicle brand can be obtained by matching the collected logo images against the vehicle brand database. Owner information can be obtained by exchanging the license plate number with vehicle management office information.

[0039] Furthermore, the step in this application, which involves multi-target tracking of the K vehicles waiting to stop based on the target traffic flow video to obtain the features of the K vehicles waiting to stop, also includes:

[0040] The first vehicle to be parked is obtained based on the K vehicles to be parked.

[0041] The first vehicle to be stopped is located in the first frame of video using a target detection algorithm to obtain the first stopping position identifier;

[0042] The first vehicle to be parked is subjected to identification feature extraction to obtain a first vehicle to be parked identification feature set, wherein the first vehicle to be parked identification feature set consists of vehicle to be parked contour features and vehicle to be parked color features;

[0043] Based on the first waiting-to-stop identification feature set, target matching is performed in the target vehicle flow video, and the position information of the first waiting-to-stop vehicle is updated in multiple rounds to obtain the first trajectory feature;

[0044] Interact with the speed measuring radar in the parking tendency monitoring area to obtain a first speed feature, and the first speed feature and the first trajectory feature constitute a first feature of a vehicle waiting to stop.

[0045] By analogy, the characteristics of the K vehicles waiting to stop are obtained.

[0046] Specifically, the methods used to obtain the features of K vehicles waiting to be parked are consistent. Therefore, this embodiment describes in detail the method of randomly selecting the first vehicle waiting to be parked from the K vehicles to obtain its features. The method uses a target detection algorithm to monitor the first vehicle waiting to be parked and locate it. Specifically, the preferred location method in this example is as follows: Preprocessing of the first frame video image of the first vehicle waiting to be parked: Preprocessing the input first frame video image, including image size adjustment and image clarity enhancement, to improve the effect of subsequent processing; Object recognition: Classifying the detected first vehicle waiting to be parked to determine its information; Target detection: Using a target detection method to locate the position point of the first vehicle waiting to be parked in the first frame video image, visualizing the location result, and marking the position of the vehicle waiting to be parked to obtain the first waiting-to-park position identifier. Feature extraction is performed on the first vehicle waiting to be parked to construct a first waiting-to-park identification feature set. The features extracted for the first vehicle waiting to be parked may include the vehicle's shape features, color features, brand features, etc., and these features are used to construct the first waiting-to-park identification feature set. The target video stream is then extracted into multiple frames. Features from the first set of vehicles to be stopped, such as their shape, color, and brand, are used to identify the vehicles waiting to stop in each frame of the multi-frame video. This ensures that the vehicle in each frame is the same as the first vehicle to be stopped. By identifying the positions of the vehicles in the multi-frame images, the position of the first vehicle to be stopped is updated multiple times. The position points of the first vehicle to be stopped in the multi-frame images are then connected to obtain the trajectory of the first vehicle to be stopped in the target traffic video. The first speed feature is obtained from roadside or roadside speed radar. The first speed feature and the first trajectory feature constitute the first feature of the vehicle to be stopped. K features of vehicles to be stopped are obtained using the same method.

[0047] Furthermore, the step in this application to synchronize the K features of vehicles waiting to park to the parking tendency analysis subnetwork to obtain K parking tendency indices also includes:

[0048] Historical data is extracted based on the parking tendency monitoring area to obtain a set of sample driving trajectories and a set of sample vehicle speeds;

[0049] The parking tendency analysis sub-network is constructed based on the knowledge graph, and the data of the parking tendency analysis sub-network is filled using the sample driving trajectory set and the sample vehicle speed set;

[0050] The first vehicle to be parked features are synchronized to the parking tendency analysis subnetwork, and feature similarity comparison is performed to obtain a set of candidate parking tendency indices.

[0051] Serialize the set of candidate parking preference indices and call to obtain the first parking preference index;

[0052] By analogy, the K parking tendency indices are obtained.

[0053] Specifically, the methods used to obtain the K parking tendency indices are consistent. Therefore, this embodiment describes in detail the method of randomly selecting the first feature of a vehicle waiting to park from the K features of vehicles waiting to park to obtain the first parking tendency index. Based on the driving trajectory and speed of vehicles that have ultimately parked in the roadside parking area in the past, a parking tendency molecular network is constructed. Each parked vehicle corresponds to its corresponding driving speed and driving trajectory. Historical data is retrieved to obtain a sample driving trajectory set and a sample vehicle speed set. The obtained driving speed and driving trajectory of the first vehicle waiting to park are then used to obtain the first parking tendency index. The data from the vehicle trajectory parking tendency analysis subnetwork is synchronized to the parking tendency analysis subnetwork. The parking tendency analysis subnetwork obtains a set of candidate parking tendency indices by performing feature comparisons and weighted summaries of driving speed and driving trajectory. For example, driving speed comparison can be performed by comparing the speed of the first vehicle to be parked with the speed of the first group of speeds; specifically, the speeds are subtracted and then divided by the speed of the first vehicle to be parked. Trajectory image comparison can use Euclidean distance to calculate image similarity. In this embodiment, the ratio of driving speed to driving trajectory is preferably set to 3:7. The candidate parking tendency indices are obtained by multiplying the comparison feature results by their respective weights. Then, the candidate parking tendency index set is obtained and serialized using the above method. The largest candidate parking tendency index is extracted as the similarity matching result, i.e., the first parking tendency index. K parking tendency indices are obtained using the same method.

[0054] Furthermore, the step in this application of obtaining the target monitored vehicle by serializing the K parking tendency indices and calling the vehicle to be parked corresponding to the largest parking tendency index also includes:

[0055] A parking probability constraint is preset, and the K parking tendency indices are traversed based on the parking probability constraint to obtain M parking tendency indices, wherein the M parking tendency indices are mapped to M vehicles waiting to park.

[0056] License plate features are extracted from the M vehicles waiting to stop to obtain M license plate numbers, where M is a positive integer;

[0057] Based on the interaction with the historical parking logs of the M parking license plate numbers, M historical parking records are obtained, wherein the historical parking records include multiple historical parking times;

[0058] M parking probability indices are calculated based on the M historical parking records;

[0059] A first weight is assigned to the parking tendency index, and a second weight is assigned to the parking probability index. Based on the first weight and the second weight, M optimized parking tendency indices are calculated.

[0060] Serialize the M optimized parking tendency indices to obtain the second target monitored vehicle;

[0061] The target monitoring vehicle is replaced with the second target monitoring vehicle.

[0062] Specifically, since parking frequencies vary across different areas—for example, the parking frequency in office buildings is definitely lower than that near tourist attractions—a parking probability constraint is pre-set. This constraint consists of two parts: a parking tendency index and historical parking frequency. M parking tendency indices are selected from K indices, each mapped to M vehicles waiting to park. Feature extraction is performed on these M vehicles to obtain M license plate numbers. These M license plate numbers are then exchanged with historical logs to obtain M historical parking records, including historical parking days. The log is composed of the license plates of all vehicles that have parked in the roadside parking area and the time of each parking. The parking probability index refers to the proportion of the number of times a vehicle parked in the roadside parking area within a certain period of time to the total number of times all vehicles parked during that period. For example, the period of time can be within a week or a month. If M represents 3 vehicles, namely A, B, and C, and the number of times these three vehicles parked in the roadside parking area in a month are 7, 5, and 6 respectively, then the parking probability index of A is: 7 ÷ (7 + 5 + 6) = 0.389. Similarly, the parking probability indices of vehicles B and C can be obtained. Finally, the parking tendency index and the parking probability index are assigned first and second weights respectively to optimize the parking tendency of vehicles waiting to park. Since the parking probability index mainly plays a regulatory role, whether or not a vehicle can park depends mainly on the current driving status of the vehicle. Therefore, the first weight is greater than the second weight. Preferably, the ratio of the first weight to the second weight in this embodiment is 8:2. M optimized parking tendency indices are obtained by multiplying the obtained parking tendency index and parking probability index by their respective weight ratios.

[0063] Furthermore, such as Figure 2 As shown, the steps in this application, prior to activating the second high-position video capture gun based on the target monitoring vehicle, also include:

[0064] Based on the license plate number of the target monitored vehicle, the system interacts with the vehicle management big data to obtain standard brand and model information and standard appearance feature information.

[0065] Based on the standard brand and model information and the standard appearance feature information, a vehicle counterfeiting risk verification is performed using the brand and model information and the appearance feature information to obtain a first verification result.

[0066] If the first verification result is successful, then the second high-position video acquisition gun is activated based on the target monitoring vehicle.

[0067] Specifically, the information of the vehicle waiting to be parked, such as the license plate number, collected by the first high-position video capture gun, is exchanged with the vehicle management office database to obtain standard brand and model information and standard appearance feature information. The standard brand and model information and standard appearance feature information refer to the feature information of the vehicle waiting to be parked recorded by the vehicle management office when the license plate number was registered. The appearance information of the vehicle waiting to be parked includes the outline, color, and brand of the vehicle. The standard brand and model information and standard appearance feature information are compared with the feature information of the current vehicle waiting to be parked to verify the risk of vehicle cloning and obtain the first verification result. If the first verification result fails, it proves that the vehicle waiting to be parked has the risk of cloning, and a first warning instruction is generated and exchanged with the traffic police team to handle the vehicle waiting to be parked. If the first verification result passes, it proves that the vehicle waiting to be parked has no risk of cloning. Then, the second high-position video capture gun is triggered based on the analysis result of the driving status of the parked vehicle.

[0068] Furthermore, the steps in this application also include:

[0069] The location coordinates of the roadside parking monitoring area are obtained interactively.

[0070] Specifically, the location coordinates refer to the roadside parking area. By determining the type of the roadside parking area, if it is a normal area, parking is allowed in that area. If the roadside parking area is in a busy area, vehicles waiting to park outside the permitted parking time are considered illegally parked.

[0071] The set of parking sections is determined based on the location coordinates of the area;

[0072] Specifically, based on the location coordinates of the monitored roadside parking area, drivers can select the parking time according to the location type. For example, the parking time in ordinary areas is all day, while the parking time in busy urban areas is 20:00-21:00.

[0073] The nearest no-parking node is determined based on the parking time identifier and the set of parking zones;

[0074] Specifically, the "near-no-parking" node refers to the time when parking is prohibited in roadside parking areas in busy urban areas. For example, if the parking time range in roadside parking areas in busy urban areas is 20:00-21:00, then the "near-no-parking" nodes in roadside parking areas in busy urban areas are 20:00 and 21:00. 21:00 is the "near-no-parking" node.

[0075] The information on the nearest no-parking node and the target vehicle fee will be sent synchronously through the payment contact channel.

[0076] Specifically, the target parking fee is calculated by the billing analysis module based on the parking time between the parking time marker and the nearest no-parking zone, and is simultaneously sent to the vehicle owner through contact channels. For example, if the target vehicle enters the roadside parking area at 20:00 and 21:00 is the nearest no-parking zone, the target vehicle should leave the roadside parking area at 21:00. The target parking fee is obtained by the billing analysis module based on the one hour of parking between 20:00 and 21:00, and can be shared with the vehicle owner information at the vehicle management office to generate a link or QR code for sending.

[0077] Example 2

[0078] Based on the same inventive concept as the aforementioned embodiment, namely, a roadside parking monitoring method based on high-position video, such as... Figure 3 As shown, this application provides a roadside parking monitoring system based on high-position video, the system comprising:

[0079] A pre-construction module for parking tendency monitoring area and roadside parking monitoring area, wherein the parking tendency monitoring area is equipped with a first high-position video acquisition gun and the roadside parking monitoring area is equipped with a second high-position video acquisition gun;

[0080] The vehicle driving dynamic analysis module is used to perform vehicle driving dynamic analysis on the traffic flow entering the parking tendency monitoring area based on the first high-position video acquisition gun, and to determine the target monitored vehicle and the target vehicle information, wherein the target vehicle information includes license plate number, brand and model, appearance characteristics and owner information.

[0081] The second high-position video acquisition gun activation module is used to activate the second high-position video acquisition gun based on the target monitoring vehicle.

[0082] The target monitoring vehicle parking time identification module is used by the second high-position video acquisition gun to identify the parking time of the target monitoring vehicle that enters the roadside parking monitoring area.

[0083] The vehicle exit analysis module is used to perform vehicle exit analysis on the target monitored vehicle and generate an exit time identifier.

[0084] A parking fee analysis module is used to preset the parking fee analysis module and send the parking time identifier, the exit time identifier and the target vehicle information to the parking fee analysis module to perform parking fee statistics and generate target parking fee information.

[0085] The fare information sending module is used to obtain payment contact channels based on the vehicle owner information interaction, and to send the target fare information based on the payment contact channels.

[0086] Furthermore, the vehicle driving dynamics analysis module of the system also includes:

[0087] The target traffic flow video acquisition unit is used to acquire video of the parking tendency monitoring area through the first high-position video acquisition gun to obtain the target traffic flow video;

[0088] The first frame video acquisition unit is used to extract video frames from the target traffic video to obtain the first frame video.

[0089] The vehicle waiting to be stopped unit is used to synchronize the first frame of video to the vehicle segmentation module to obtain K vehicles waiting to be stopped, wherein the K vehicles waiting to be stopped correspond to K vehicle code identifiers, and K is a positive integer;

[0090] The vehicle tracking unit is used to perform multi-target tracking of the K vehicles waiting to stop based on the target traffic flow video, and obtain K features of the vehicles waiting to stop, wherein the features of the vehicles waiting to stop include speed features and trajectory features;

[0091] The target monitoring vehicle acquisition unit is used to synchronize the features of the K vehicles waiting to be parked to the parking tendency analysis sub-network, obtain K parking tendency indices, and obtain the target monitoring vehicle by serializing the K parking tendency indices and calling the vehicle waiting to be parked corresponding to the largest parking tendency index.

[0092] The vehicle key feature extraction unit is used to extract key features of the target monitored vehicle to obtain the target vehicle information.

[0093] Furthermore, the vehicle tracking unit of the system also includes:

[0094] The first waiting-to-park vehicle acquisition unit is used to extract and obtain the first waiting-to-park vehicle based on the K waiting-to-park vehicles;

[0095] The first waiting-to-stop location identifier acquisition unit is used to locate the first waiting-to-stop vehicle in the first frame of video using a target detection algorithm to obtain the first waiting-to-stop location identifier;

[0096] The first waiting-to-stop identification feature set acquisition unit is used to extract identification features from the first waiting-to-stop vehicle to obtain the first waiting-to-stop identification feature set, wherein the first waiting-to-stop identification feature set consists of waiting-to-stop contour features and waiting-to-stop color features;

[0097] The first trajectory feature acquisition unit is used to perform target matching in the target vehicle flow video based on the first stop identification feature set, update the position information of the first stop vehicle in multiple rounds, and obtain the first trajectory feature.

[0098] The first vehicle waiting to stop feature acquisition unit is used to interact with the speed measuring radar of the parking tendency monitoring area to obtain a first speed feature, and the first speed feature and the first trajectory feature constitute the first vehicle waiting to stop feature.

[0099] K features acquisition units for waiting vehicles are used to acquire the features of the K waiting vehicles.

[0100] Furthermore, the first vehicle feature acquisition unit of the system also includes:

[0101] The historical data extraction unit is used to extract historical data based on the parking tendency monitoring area to obtain a set of sample driving trajectories and a set of sample vehicle speeds.

[0102] The parking tendency analysis subnetwork construction unit is used to construct the parking tendency analysis subnetwork based on the knowledge graph, and to fill the parking tendency analysis subnetwork with the sample driving trajectory set and the sample vehicle speed set;

[0103] The candidate parking tendency index set acquisition unit is used to synchronize the features of the first vehicle to be parked with the parking tendency analysis sub-network, perform feature similarity comparison, and obtain the candidate parking tendency index set.

[0104] The first parking tendency index acquisition unit is used to serialize the set of candidate parking tendency indices and call to obtain the first parking tendency index;

[0105] K parking tendency index acquisition units are used to obtain the K parking tendency indexes.

[0106] Furthermore, the vehicle driving dynamics analysis module of the system also includes:

[0107] A parking probability constraint preset unit is used to preset parking probability constraints and traverse the K parking tendency indices based on the parking probability constraints to obtain M parking tendency indices, wherein the M parking tendency indices are mapped to M vehicles waiting to park.

[0108] The unit for obtaining M parking license plate numbers is used to extract license plate features from the M vehicles waiting to park and obtain M parking license plate numbers, where M is a positive integer;

[0109] The M historical parking record acquisition unit is used to obtain M historical parking records by interacting with the historical parking log based on the M parking license plate numbers, wherein the historical parking records include multiple historical parking times;

[0110] M parking probability index acquisition units are used to calculate M parking probability indices based on the M historical parking records;

[0111] M optimized parking tendency index acquisition units are used to assign a first weight to the parking tendency index and a second weight to the parking probability index, and to calculate M optimized parking tendency indices based on the first weight and the second weight.

[0112] The second target monitoring vehicle acquisition unit is used to serialize the M optimized parking tendency indices to obtain the second target monitoring vehicle.

[0113] A target monitoring vehicle replacement unit is used to replace the target monitoring vehicle with the second target monitoring vehicle.

[0114] Furthermore, the system's second high-position video acquisition gun activation module also includes:

[0115] The standard brand model information and standard appearance feature information acquisition unit is used to obtain standard brand model information and standard appearance feature information by interacting with vehicle management big data based on the license plate number of the target monitored vehicle;

[0116] The first verification result obtaining unit is used to perform vehicle license plate counterfeiting risk verification based on the standard brand and model information and the standard appearance feature information, and to obtain the first verification result.

[0117] The second high-position video acquisition gun activation unit is used to activate the second high-position video acquisition gun based on the target monitoring vehicle if the first verification result is a successful verification.

Claims

1. A method for monitoring roadside parking based on high-position video, characterized in that, The method includes: A parking tendency monitoring area and a roadside parking monitoring area are pre-constructed, wherein the parking tendency monitoring area is equipped with a first high-position video acquisition gun and the roadside parking monitoring area is equipped with a second high-position video acquisition gun; Based on the first high-position video acquisition gun, the vehicle dynamics of the traffic entering the parking tendency monitoring area are analyzed to determine the target monitored vehicle and the target vehicle information, wherein the target vehicle information includes license plate number, brand and model, appearance characteristics and owner information; this step includes acquiring video of the target traffic flow by using the first high-position video acquisition gun to capture video of the parking tendency monitoring area. The target traffic flow video is processed to extract video frames to obtain the first video frame; The first frame of video is synchronized to the vehicle segmentation module to obtain K vehicles waiting to stop, wherein the K vehicles waiting to stop correspond to K vehicle code identifiers, and K is a positive integer; Based on the target traffic video, multi-target tracking of the K vehicles waiting to stop is performed to obtain K features of the vehicles waiting to stop, wherein the features of the vehicles waiting to stop include speed features and trajectory features; The features of the K vehicles waiting to be parked are synchronized to the parking tendency analysis subnetwork to obtain K parking tendency indices. The target monitored vehicle is obtained by serializing the K parking tendency indices and calling the vehicle waiting to be parked corresponding to the largest parking tendency index. Key features of the target monitored vehicle are extracted to obtain the target vehicle information; The second high-position video acquisition gun is activated based on the target monitored vehicle; The second high-position video capture camera marks the parking time of the target monitored vehicle that enters the roadside parking monitoring area; Perform vehicle departure analysis on the target monitored vehicle and generate departure time identifier; A parking fee analysis module is preset, and the parking time identifier, the exit time identifier and the target vehicle information are sent to the parking fee analysis module to perform fee statistics and generate target fee information. The payment contact channel is obtained based on the vehicle owner information interaction, and the target fare information is sent based on the payment contact channel; The method further includes synchronizing the characteristics of the K vehicles waiting to park to the parking tendency analysis subnetwork to obtain K parking tendency indices. Historical data is extracted based on the parking tendency monitoring area to obtain a set of sample driving trajectories and a set of sample vehicle speeds; The parking tendency analysis sub-network is constructed based on the knowledge graph, and the data of the parking tendency analysis sub-network is filled using the sample driving trajectory set and the sample vehicle speed set; Among the K vehicles waiting to park, the first feature of the vehicle waiting to park is randomly selected, and the first feature of the vehicle waiting to park is synchronized to the parking tendency analysis sub-network for feature similarity comparison to obtain a set of candidate parking tendency indices. Serialize the set of candidate parking preference indices and call to obtain the first parking preference index; By analogy, the K parking tendency indices are obtained.

2. The method as described in claim 1, characterized in that, Based on the target traffic flow video, multi-target tracking of the K vehicles waiting to stop is performed to obtain the features of the K vehicles waiting to stop. The method further includes: The first vehicle to be parked is obtained based on the K vehicles to be parked. The first vehicle to be stopped is located in the first frame of video using a target detection algorithm to obtain the first stopping position identifier; The first vehicle to be parked is subjected to identification feature extraction to obtain a first vehicle to be parked identification feature set, wherein the first vehicle to be parked identification feature set consists of vehicle to be parked contour features and vehicle to be parked color features; Based on the first waiting-to-stop identification feature set, target matching is performed in the target traffic video, and the position information of the first waiting-to-stop vehicle is updated in multiple rounds to obtain the first trajectory feature; Interact with the speed measuring radar in the parking tendency monitoring area to obtain a first speed feature, and the first speed feature and the first trajectory feature constitute a first feature of a vehicle waiting to stop. By analogy, the characteristics of the K vehicles waiting to stop are obtained.

3. The method as described in claim 1, characterized in that, The method further includes obtaining the target monitored vehicle by serializing the K parking tendency indices and calling the vehicle to be parked corresponding to the largest parking tendency index. A parking probability constraint is preset, and the K parking tendency indices are traversed based on the parking probability constraint to obtain M parking tendency indices, wherein the M parking tendency indices are mapped to M vehicles waiting to park. License plate features are extracted from the M vehicles waiting to stop to obtain M license plate numbers, where M is a positive integer; Based on the interaction with the historical parking logs of the M parking license plate numbers, M historical parking records are obtained, wherein the historical parking records include multiple historical parking times; M parking probability indices are calculated based on the M historical parking records; A first weight is assigned to the parking tendency index, and a second weight is assigned to the parking probability index. Based on the first weight and the second weight, M optimized parking tendency indices are calculated. Serialize the M optimized parking tendency indices to obtain the second target monitored vehicle; The target monitoring vehicle is replaced with the second target monitoring vehicle.

4. The method as described in claim 1, characterized in that, Based on the activation of the second high-position video acquisition gun by the target monitoring vehicle, the method further includes: Based on the license plate number of the target monitored vehicle, the system interacts with the vehicle management big data to obtain standard brand and model information and standard appearance feature information. Based on the standard brand and model information and the standard appearance feature information, a vehicle counterfeiting risk verification is performed using the brand and model information and the appearance feature information to obtain a first verification result. If the first verification result is successful, then the second high-position video acquisition gun is activated based on the target monitoring vehicle.

5. The method as described in claim 1, characterized in that, The method further includes: The location coordinates of the roadside parking monitoring area are obtained interactively. The set of parking sections is determined based on the location coordinates of the area; The nearest no-parking node is determined based on the parking time identifier and the set of parking zones; The information on the nearest no-parking node and the target vehicle fee will be sent synchronously through the payment contact channel.

6. A roadside parking monitoring system based on high-position video, used to execute the method of claim 1, characterized in that, The system includes: A pre-construction module for parking tendency monitoring area and roadside parking monitoring area, wherein the parking tendency monitoring area is equipped with a first high-position video acquisition gun and the roadside parking monitoring area is equipped with a second high-position video acquisition gun; The vehicle driving dynamic analysis module is used to perform vehicle driving dynamic analysis on the traffic flow entering the parking tendency monitoring area based on the first high-position video acquisition gun, and to determine the target monitored vehicle and the target vehicle information, wherein the target vehicle information includes license plate number, brand and model, appearance characteristics and owner information. The second high-position video acquisition gun activation module is used to activate the second high-position video acquisition gun based on the target monitoring vehicle. The target monitoring vehicle parking time identification module is used by the second high-position video acquisition gun to identify the parking time of the target monitoring vehicle that enters the roadside parking monitoring area. The vehicle exit analysis module is used to perform vehicle exit analysis on the target monitored vehicle and generate an exit time identifier. A parking fee analysis module is used to preset the parking fee analysis module and send the parking time identifier, the exit time identifier and the target vehicle information to the parking fee analysis module to perform parking fee statistics and generate target parking fee information. The fare information sending module is used to obtain payment contact channels based on the vehicle owner information interaction, and to send the target fare information based on the payment contact channels.

Images