Parking management method and system based on high-bit video

Abstract

The application discloses a parking management method and system based on high-bit video and belongs to the field of intelligent traffic management, wherein the method comprises the following steps: collecting a video of a supervision area to determine a parking space monitoring video; feeding the parking space monitoring video to a background processor to analyze and determine parking features frame by frame, wherein the parking features are provided with time sequence marks; generating a parking data chain based on the parking features, wherein the parking data chain is provided with illegal information marks; determining a regional distribution map based on the parking space monitoring video, identifying vacant parking spaces, generating a regional parking space real-time map, and the regional parking space real-time map has real-time updating performance; and performing visual display and parking management based on a central management system according to the regional parking space real-time map and the parking data chain. The application solves the technical problems of low parking management efficiency and low parking resource utilization rate in the prior art, and achieves the technical effects of improving the parking management efficiency and the parking resource utilization rate.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent traffic management, in particular to a parking management method and system based on high-position video. BACKGROUND

[0002] With the acceleration of urbanization and the increase of travel needs, the number of parking lots is increasing, and parking management is facing the problems of space shortage and low efficiency. The existing parking management system mainly uses door magnetic, parking sensor and other devices to collect parking information for parking management, which is difficult to accurately capture the information of vehicle entering and leaving, and the management efficiency is low. At the same time, based on the single management model, it is impossible to identify and manage the historical parking behavior. SUMMARY

[0003] The present application provides a parking management method and system based on high-position video, aiming at solving the technical problems of low parking management efficiency and low parking resource utilization in the prior art.

[0004] In view of the above problems, the present application provides a parking management method and system based on high-position video.

[0005] The first aspect of the present application provides a parking management method based on high-position video, which comprises: video acquisition of a supervision area to determine a parking space monitoring video, the supervision area is configured with at least one monitoring device, each monitoring device corresponds to a parking space coverage domain; feeding the parking space monitoring video to a background processor to determine parking features frame by frame, the parking features have time sequence identification; based on the parking features, generating a parking data chain, the parking data chain identifies illegal information; based on the parking space monitoring video, determining a regional distribution map and identifying vacant parking spaces to generate a regional parking space real-time map, the regional parking space real-time map has real-time updating property; according to the regional parking space real-time map and the parking data chain, visual display and parking management are carried out based on a central management system.

[0006] Another aspect of the present application provides a parking management system based on high-position video, which comprises: a monitoring video determination module for video acquisition of a supervision area to determine a parking space monitoring video, the supervision area is configured with at least one monitoring device, each monitoring device corresponds to a parking space coverage domain; a parking feature determination module for feeding the parking space monitoring video to a background processor to determine parking features frame by frame, the parking features have time sequence identification; a parking data chain module for generating a parking data chain based on the parking features, the parking data chain identifies illegal information; a parking space real-time map module for determining a regional distribution map based on the parking space monitoring video and identifying vacant parking spaces to generate a regional parking space real-time map, the regional parking space real-time map has real-time updating property; a visual parking management module for visual display and parking management based on a central management system according to the regional parking space real-time map and the parking data chain.

[0007] The one or more technical solutions provided in the application have at least the following technical effects or advantages:

[0008] Due to the video collection of the supervision area, the parking space monitoring video is determined; the parking space monitoring video is fed back to the background processor, the parking features are determined by frame-by-frame analysis, the parking features have time sequence identification; based on the parking features, the parking data chain is generated, the parking data chain has illegal information; based on the parking space monitoring video, the regional distribution map is determined and the vacant parking space is identified, the regional parking live map is generated, and the regional parking live map has real-time updating; according to the regional parking live map and the parking data chain, the central management system is used for visual display and parking management, which solves the technical problems of low parking management efficiency and low parking resource utilization in the prior art, and achieves the technical effects of improving the parking management efficiency and the parking resource utilization.

[0009] The above description is only a summary of the technical solutions of the application, in order to more clearly understand the technical means of the application, the application can be implemented according to the content of the specification, and in order to make the above and other purposes, characteristics and advantages of the application more obvious and easy to understand, the following specific embodiments of the application are described. BRIEF DESCRIPTION OF DRAWINGS

[0010] Figure 1 A possible flowchart of the parking management method based on high-level video is provided for the embodiments of the application;

[0011] Figure 2 A possible flowchart of determining parking features in the parking management method based on high-level video is provided for the embodiments of the application;

[0012] Figure 3 A possible flowchart of generating a parking data chain in the parking management method based on high-level video is provided for the embodiments of the application;

[0013] Figure 4 A possible structure diagram of the parking management system based on high-level video is provided for the embodiments of the application.

[0014] Explanation of reference signs: monitoring video determination module 11, parking feature determination module 12, parking data chain module 13, parking live map module 14, and visual parking management module 15. DETAILED DESCRIPTION

[0015] The general idea of the technical solutions provided in the application is as follows:

[0016] This application provides a parking management method and system based on high-position video surveillance. By configuring video monitoring equipment in the monitored area, each parking space is monitored comprehensively to obtain accurate parking space usage information and vehicle entry and exit information. The collected video information is used as the basis of the management system. After information processing and analysis, a real-time data chain of parking space status and a real-time map of parking spaces in the area are generated. Based on the real-time parking space status data, visualized parking display and refined management are implemented.

[0017] After introducing the basic principles of this application, various non-limiting embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0018] Example 1

[0019] like Figure 1 As shown in the figure, this application provides a parking management method based on high-view video, the method including:

[0020] Step S1000: Collect video of the monitored area to determine the parking space monitoring video. The monitored area is equipped with at least one monitoring device, and each monitoring device corresponds to a parking space coverage area.

[0021] Specifically, for areas requiring parking space management, i.e., monitored areas, surveillance videos are collected to determine the videos used to monitor the status of parking spaces, which serve as the parking space monitoring videos. At least one monitoring device is deployed in each monitored area, with each device corresponding to one parking space, i.e., the parking space coverage area.

[0022] The monitored area includes parking lots, highway sections, or urban roads where vehicle parking management is required. Video surveillance equipment, such as cameras and video capture devices, is deployed at key locations within the monitored area, such as entrances, exits, turning points, and areas with dense parking spaces. Each monitoring device monitors a specific parking space coverage area, tracking the parking spaces and vehicles within that area. The video surveillance equipment is connected to a video capture terminal, such as a video capture card, to acquire and transmit the video signals obtained by the monitoring equipment in real time. The video capture terminal encodes, decodes, and formats the acquired video signals, selecting the video used to monitor the parking space status to generate parking space monitoring video.

[0023] By deploying surveillance equipment in the regulated area to acquire surveillance video and identifying videos for monitoring parking space status, video data can be provided to support subsequent analysis of parking space conditions and vehicle parking, which is beneficial for parking space location and vehicle location.

[0024] Step S2000: Feed the parking space monitoring video back to the background processor, analyze it frame by frame to determine the parking features, and the parking features are marked with time sequence.

[0025] Specifically, the generated parking space monitoring video is sent to a background processor in the background management system. The background processor analyzes the parking space monitoring video frame by frame to determine the characteristic information of the vehicle parking, and the parking characteristics contain time sequence identifiers for identifying the time information of the characteristic generation.

[0026] Firstly, the parking space monitoring video is sent to the background processor in real time through wired or wireless means for video analysis and feature recognition; secondly, the background processor decodes and formats the received parking space monitoring video to extract the image frame sequence in the video; then, the image frame sequence is analyzed frame by frame, and edge detection, image segmentation and other methods are used to extract the target area in each image frame, and then classification or matching methods are used to determine whether the target area is a vehicle; then, the detected vehicle target is tracked and positioned to obtain its parking position, posture and other information in the parking space, and the time sequence identifier of the parking characteristic is generated according to the video acquisition time. At the same time, different parking characteristics are determined for different parking positions and postures, for example, parallel parking / vertical parking corresponds to different position information; forward parking / reverse parking corresponds to different posture information, etc. The obtained parking characteristics are stored and managed for subsequent retrieval and use. The storage content includes the parking characteristics and their time sequence identifiers for identifying the time information of the characteristic acquisition.

[0027] By analyzing the parking space monitoring video frame by frame, the vehicle parking characteristics and their time sequence identifiers are obtained, which provides a basis for accurately obtaining the parking information of the vehicle in the parking space and subsequent parking management and retrieval.

[0028] Step S3000: based on the parking characteristics, a parking data chain is generated, and the parking data chain identifies the violation information;

[0029] Specifically, the parking data chain is an ordered data set composed of multiple nodes, each node corresponding to a parking characteristic, and the nodes are connected through time sequence relationship. The parking data chain represents the complete parking process of the vehicle in the parking space, including all information of the vehicle entering, parking and leaving the parking space. The violation information refers to the information that violates the parking rules in the vehicle parking process, such as overtime parking, reverse parking, occupying the disabled parking space, etc. The violation information is realized by adding identifiers in the corresponding nodes of the parking data chain.

[0030] Firstly, the acquired parking features are sorted according to the feature acquisition time to obtain the time sequence of the vehicle parking. Then, according to the time sequence, the adjacent parking features representing the vehicle parking in the same parking space are connected to form a node, and each node corresponds to the complete parking information of the vehicle in a parking space. Then, according to the time sequence relationship between the nodes, the nodes are connected in time sequence to form a parking data chain representing the complete parking process of the vehicle. In each node of the parking data chain, the violation information representing the violation events generated in the period of the node is added, for example, the overtime parking adds the overtime parking identifier; the reverse parking adds the reverse parking identifier, etc.

[0031] By connecting the parking features in the same parking space and at the same time to form a node, and then connecting the nodes in time sequence to form a parking data chain representing the complete parking process of the vehicle, the management of the vehicle parking information is realized. At the same time, the violation information is marked in the data chain, the management of the whole parking process of the vehicle and the identification of the illegal parking are realized, which provides data support for accurate and scientific management of vehicle parking.

[0032] Step S4000: Based on the parking space monitoring video, the region distribution map is determined and the vacant parking space is identified, and the region parking space real-time map is generated, which has real-time updating property;

[0033] Specifically, first, the region distribution information is extracted based on the video picture for monitoring the parking space, and the region distribution map is generated. The region distribution map detects the region contour in the picture through video analysis technology, identifies the spatial relationship between the regions, and generates a topological connection relationship map between the regions, which is used to identify the spatial distribution and mutual relationship of each parking space in the supervision region. Then, based on the region distribution map, the vacant parking spaces in each region are identified and marked. Whether a vehicle is parked on the parking space is determined through the environment modeling in the video, if no vehicle is detected on the parking space, the parking space is marked as a vacant parking space. Then, the identification result of the vacant parking space is combined with the connection relationship of the region distribution map to generate a region parking space real-time map. The region parking space real-time map identifies the vacancy and occupancy of each parking space in the supervision region, and the map has real-time updating property, which can dynamically perceive the change of the parking space state in the region and provide real-time reference information for parking management and parking guidance.

[0034] The region distribution map is determined and the vacant parking space is identified through video monitoring to generate a region parking space real-time map, which realizes the perception and judgment of the parking space state in the supervision region and provides information basis for parking management.

[0035] Step S5000: According to the region parking space real-time map and the parking data chain, the central management system is used for visual display and parking management.

[0036] Specifically, first, two information sources of regional parking live map and parking data chain need to be obtained. The regional parking live map identifies the vacancy and occupancy of each parking space in the supervision area, and can reflect the parking space status in the area in real time. The parking data chain contains dynamic information in the vehicle parking process, such as the time of vehicle entering and leaving the supervision area, the trajectory of vehicle moving in the area, etc., which can restore the parking process of the vehicle.

[0037] Then, the management and display of information are realized based on the central management system. As the core of the supervision system, the central management system is responsible for information reception, processing and transmission. After receiving real-time feedback of the two information sources of regional parking live map and parking data chain, the system stores and manages the information in the database. Then, the parking space status map, the animation simulation of vehicle moving trajectory, and the page display of vehicle basic information are displayed in a visual display mode. In addition to information display, the central management system manages the vehicle parking according to the information source, such as allocating and guiding the parking space for the vehicle based on the time of vehicle entering and leaving and the real-time information of vacant parking space, implementing parking management; if there is illegal parking, the system manages and reminds the relevant vehicle through information matching.

[0038] By establishing the central management system, the centralized management and comprehensive use of information such as regional parking live map and parking data chain are realized, and the related information is displayed in a visual way, which is convenient for supervision and command. The intelligent parking management based on video monitoring is realized, and the parking management efficiency and parking resource utilization rate are improved.

[0039] Further, the embodiments of the application also include:

[0040] Step S2100: frame-by-frame segmentation of the parking space monitoring video is performed to determine a plurality of video frames;

[0041] Step S2200: based on the plurality of video frames, edge detection is performed to extract a starting frame image, an ending frame image and a key frame image, and the image edge displacement value of adjacent two frames in the key frame image meets a threshold value standard;

[0042] Step S2300: image segmentation and feature recognition are performed on the starting frame image, the ending frame image and the key frame image to determine parking features.

[0043] Specifically, first, a video stream for monitoring a parking space is collected and segmented. The video stream is a dynamic picture composed of continuous static images, each static image corresponds to a video frame, and a plurality of video frames are extracted by segmenting the video stream, which record the picture information of the parking space at consecutive time points. After obtaining a plurality of video frames, edge information in each frame image is detected by edge detection technology to determine the change of edges between adjacent two frame images. If the edge displacement value is large, it indicates that the object in the region between the two frame images moves quickly, and the frame image is not suitable as a key frame image. Only when the edge displacement value of adjacent two frame images meets the predetermined threshold standard, the frame image can be determined as a key frame image. At the same time, the selection of the starting frame image and the ending frame image also depends on the result of edge detection. The starting frame image is captured before the object starts to move, and the ending frame image is captured after the object stops moving. The selection of the two is determined according to the change trend of the edge displacement value.

[0044] Then, the starting frame image, the ending frame image and the key frame image are segmented by detecting the boundary of the object and identifying the type of the object, and the image is divided into a plurality of groups, each group corresponding to an identified target in the parking process, such as a vehicle body, a license plate, etc. Then, a dynamic feature recognition channel and a static feature recognition channel are constructed, the former extracts features from dynamic information of the image, and the latter extracts features from static components of the image. Feature recognition is performed on each group of segmented images through the constructed feature recognition channel. The characteristics and state information of the vehicle are comprehensively judged, and the parking features are finally determined.

[0045] Through video stream processing, image segmentation and feature recognition, etc. Technical means, the extraction and judgment of the parking related information in the monitoring video are realized, the parking features are output, and the basic information acquisition of the parking space monitoring and parking management is completed.

[0046] Further, the embodiments of the application also include:

[0047] Step S2310: determining an identified target, performing image segmentation on the starting frame image, the ending frame image and the key frame image based on the identified target, respectively, to determine a plurality of groups of segmented images, wherein each group of segmented images corresponds to an identified target;

[0048] Step S2320: building a feature recognition model, including a dynamic feature recognition channel and a static feature recognition channel arranged in parallel;

[0049] Step S2330: inputting the plurality of groups of segmented images into the feature recognition model, and outputting the parking features.

[0050] Specifically, first, according to the requirements of parking management, analyze the information that needs to be extracted from the image, such as license plate number, vehicle body area, etc., which correspond to different identification targets; determine the priority of image recognition, different identification targets have different importance to parking management, and the feature information corresponding to the high priority identification target is more critical to the system judgment; then, select an image segmentation method, such as color-based segmentation for license plate area and edge-based segmentation for vehicle body area; adopt the selected segmentation method to segment the starting frame image, the ending frame image and the key frame image to obtain multiple groups of segmented images. At the same time, each group of segmented images is labeled, and the label content includes the category to which the image belongs, the identification target, the acquisition time and other information.

[0051] Then, the channel adopts a convolutional neural network as a deep learning model, the input is multiple continuous segmented images, and the output is driving trajectory, driving speed and other information, a dynamic feature recognition channel is constructed, which is used to extract dynamic information from image sequences and judge the driving state of the vehicle; the channel support vector machine is a learning model, the input is a single segmented image, and the output is vehicle brand, vehicle body color, violation feature and other information, a static feature recognition channel is constructed, which is used to extract static information from a single segmented image and judge the basic features of the vehicle. The dynamic feature recognition channel and the static feature recognition channel are arranged in parallel. The outputs of the two channels are combined to realize comprehensive judgment of the vehicle state.

[0052] Next, the multiple groups of segmented images are preprocessed, such as image rotation, denoising and normalization, to improve the image quality and meet the input requirements of the model; the preprocessed multiple groups of segmented images are used as input samples, which include image information, image segmentation target, acquisition time and other information; the constructed input samples are input into the dynamic feature recognition channel or the static feature recognition channel for feature extraction and recognition. The dynamic feature recognition channel outputs dynamic information such as vehicle driving trajectory; the static feature recognition channel outputs static information such as license plate number and vehicle body color to form the feature recognition result; the results of the feature recognition channels are comprehensively judged to obtain the parking features of the vehicle, such as judging whether it is a violation parking or a normal parking.

[0053] Further, as shown in Figure 2 the embodiment of the present application further comprises:

[0054] Step S2331: input the multiple groups of segmented images into the feature recognition model;

[0055] Step S2332: based on the dynamic feature recognition channel, output the parking state, the parking state includes the driving trajectory;

[0056] Step S2333: based on the static feature recognition model, output the feature state, the feature state includes the violation feature and the vehicle basic information;

[0057] Step S2334: Determine the parking feature based on the parking state and the feature state.

[0058] Specifically, after acquiring multiple sets of segmented images, these images are input into a constructed feature recognition model for feature extraction and recognition. The feature recognition model includes a dynamic feature recognition channel and a static feature recognition channel. The former mainly analyzes the dynamic information of the image, while the latter mainly analyzes the static information of the image.

[0059] The dynamic feature recognition channel extracts vehicle motion features from image sequences to determine the vehicle's driving state and trajectory. For example, it tracks and judges changes in vehicle direction and speed from the temporal and spatial information of multiple segmented images, outputting a trajectory map of the vehicle's movement – ​​this represents the driving trajectory information in a parking state. The static feature recognition channel analyzes a single segmented image, extracting vehicle features from the image's static elements. For example, it identifies the license plate number from a segmented image of the license plate area; and determines whether there are features indicating illegal parking from a segmented image of the vehicle body area. This information constitutes the feature state, including both basic vehicle information and the presence of illegal parking features.

[0060] Then, the parking status is analyzed to examine the vehicle's trajectory and speed changes, determining whether its driving status is normal and whether there is any abnormal driving behavior; the feature status is also analyzed to determine if there are any anomalies in the vehicle's parking position. The analysis results of dynamic and static information are combined to obtain the vehicle's parking characteristics.

[0061] By constructing dynamic feature recognition channels and static feature recognition channels, the dynamic and static information of images is analyzed to determine the parking status and feature status of vehicles, and parking features are determined accordingly. This enables efficient and accurate judgment of vehicle status in parking lots, thereby improving parking management efficiency.

[0062] Furthermore, such as Figure 3 As shown, embodiments of this application also include:

[0063] Step S3100: Based on the driving trajectory, determine the parking time zone, which is the time interval between entering and exiting the parking space coverage area of ​​the monitoring equipment;

[0064] Step S3200: Perform time-series node segmentation on the parking time zone to determine the link time-series nodes, wherein the time intervals of adjacent link time-series nodes are not consistent;

[0065] Step S3300: Traverse the link time sequence nodes, match the parking features and node identifiers, and generate the parking data chain.

[0066] Specifically, first, the driving track information of the vehicle is acquired, the driving track records the time points when the vehicle enters and exits the parking area covered by the monitoring device, and is used to determine the driving state of the vehicle during this period. Then, according to the vehicle entry and exit time points recorded in the driving track, the parking time zone is determined, which represents the time interval during which the vehicle is in the monitoring device coverage domain and during which the vehicle is active in the high-position video detection area.

[0067] After obtaining the parking time zone, it is finely divided to determine a plurality of time sequence nodes, each of which represents a time point in the parking time zone, and the time interval between adjacent two time sequence nodes can be inconsistent. Among them, the time interval between two adjacent nodes should be within the range where the video captures obvious picture changes, so as to determine the link time sequence node. After obtaining the link time sequence node, each time sequence node is traversed, and the monitoring video picture at the corresponding time point is captured to match and identify the parking features therein. The parking features matched by each time sequence node are bound to the time information of the node to generate a data unit. A plurality of data units are connected in time sequence to form a parking data chain, which records the detailed state and activity information of the vehicle at each time sequence node in the parking time zone.

[0068] By acquiring the driving track of the vehicle, determining the parking time zone and the link time sequence node, and capturing the picture at each node to match the parking features, a parking data chain is finally generated, which provides a data basis for improving the parking management efficiency.

[0069] Further, the embodiments of the present application also include:

[0070] Step S5100: generating pre-warning information for the violation features based on the parking data chain to alarm the vehicle violation;

[0071] Step S5200: if the violation features meet the feature threshold, performing background record processing;

[0072] Step S5300: based on the area parking live map, visually displaying and guiding the terminal display device at the parking site.

[0073] Specifically, first, the parking data chain information is acquired, and the parking data chain records detailed activity information of the vehicle in the parking area, including driving track, stay time, violation features and the like; then, the information in the parking data chain is analyzed to determine whether there is a violation feature that violates the management regulations. If the violation feature is detected, such as shielding the license plate, occupying the regular parking space for a long time and the like, corresponding early warning information is generated to perform the violation warning. The early warning information is sent to the vehicle owner through a display screen, a short message, a voice and the like to prompt the vehicle owner to correct the violation behavior. Meanwhile, the early warning information is also sent to the management personnel for supervision and processing. After the violation feature is detected and the vehicle is warned, the severity of the violation feature is determined. If the violation feature meets a predetermined feature threshold, a background record processing is performed. The information of the violation vehicle, the violation feature, the punishment measure and the like are recorded for the accumulation and tracking processing of the management data. When the violation feature reaches a certain number of times, more severe punishment means is taken for the vehicle to strengthen the management effect.

[0074] The vacancy of each parking space in the area is recorded in the area parking live graph, which can be used for guiding display of the parking space. A terminal display device such as a display screen is arranged at the parking site to collect and display the area parking live graph information in real time. The vehicle owner can visually check the vacancy of the parking space in the area, select the vacant parking space for parking, facilitate the vehicle owner to find the vacant parking space, avoid the management difficulty caused by random parking, and simultaneously perform the parking space management and allocation according to the guiding information to improve the management efficiency.

[0075] Further, the embodiment of the application further includes:

[0076] Step S6100: setting a directional operation and maintenance period to perform operation and maintenance management of the monitoring device;

[0077] Step S6200: calling the abnormal supervision record in the central management system within a predetermined time interval and performing frequency statistics to determine the abnormal record frequency. If the abnormal record frequency is greater than or equal to a frequency threshold, a temporary operation and maintenance node is generated;

[0078] Step S6300: adding the temporary operation and maintenance node into the directional operation and maintenance period.

[0079] Specifically, first, according to the technical parameters of the device and management requirements and other factors, the directional operation and maintenance period of the monitoring device is set, and the time interval for operation and maintenance management of the monitoring device is specified. Whenever a maintenance period is reached, the device is maintained and checked to ensure the normal operation of the device and timely discovery and repair of abnormal problems. In the process of device operation, the central management system will record various abnormal supervision events, such as device failure, software error, etc. information. Periodically call these abnormal records and conduct statistical analysis to determine the frequency of a certain abnormal event within a certain time interval. If the frequency of a certain abnormal event reaches or exceeds the frequency threshold set in advance, it indicates that the event is more serious, and a temporary operation and maintenance node is generated to manage and handle the abnormal event. After generating the temporary operation and maintenance node, it is integrated into the original directional operation and maintenance period and is managed together.

[0080] By setting the operation and maintenance period, calling the abnormal supervision records and generating the temporary node, the abnormal management of the monitoring device is realized, the refinement of the parking management is improved, and the management effect is improved.

[0081] In summary, the parking management method based on high-level video provided by the embodiments of the present application has the following technical effects:

[0082] Video collection is performed on the supervision area to determine the parking space monitoring video. The supervision area is configured with at least one monitoring device, and each monitoring device corresponds to a parking space coverage domain. By obtaining accurate parking space usage information, a basis is provided for information analysis and management. The parking space monitoring video is fed back to the background processor, and parking features are determined by frame-by-frame analysis. The parking features have time sequence identifiers, providing data for generating data chains and live maps. Based on the parking features, parking data chains are generated, and the parking data chains identify illegal information, providing a basis for management system decision-making. Based on the parking space monitoring video, a regional distribution map is determined and the vacant parking spaces are identified, and a regional parking live map is generated. The regional parking live map has real-time updating property and depicts the parking live in the region, providing real-time basis for refined management. According to the regional parking live map and the parking data chain, visual display and parking management are performed based on the central management system, achieving the technical effects of improving parking management efficiency and parking resource utilization rate.

[0083] Embodiment Two

[0084] Based on the same inventive concept as the parking management method based on high-level video in the foregoing embodiments, as shown in Figure 4 The embodiments of the present application provide a parking management system based on high-level video, which comprises:

[0085] The monitoring video determination module 11 is configured to collect video of the supervision area and determine a parking space monitoring video. The supervision area is configured with at least one monitoring device, and each monitoring device corresponds to a parking space coverage domain.

[0086] a parking feature determination module 12 configured to feed the parking space monitoring video to a background processor, analyze and determine parking features frame by frame, and mark the parking features with time sequence;

[0087] a parking data chain module 13 configured to generate a parking data chain based on the parking features, and mark the parking data chain with violation information;

[0088] a parking live map module 14 configured to determine a regional distribution map and identify vacant parking spaces based on the parking space monitoring video, and generate a regional parking live map, which is real-time updated;

[0089] a visual parking management module 15 configured to display and manage parking based on the regional parking live map and the parking data chain according to a central management system.

[0090] Further, the parking feature determination module 12 comprises the following execution steps:

[0091] frame-by-frame segmentation of the parking space monitoring video to determine a plurality of video frames;

[0092] edge detection based on the plurality of video frames to extract a start frame image, an end frame image, and a key frame image, wherein the image edge displacement value of adjacent two frames in the key frame image meets a threshold value standard;

[0093] image segmentation and feature recognition of the start frame image, the end frame image, and the key frame image to determine parking features.

[0094] Further, the parking feature determination module 12 further comprises the following execution steps:

[0095] determination of identification targets, image segmentation of the start frame image, the end frame image, and the key frame image based on the identification targets respectively to determine a plurality of sets of segmented images, wherein one set of segmented images corresponds to one identification target;

[0096] building a feature recognition model comprising a dynamic feature recognition channel and a static feature recognition channel arranged in parallel;

[0097] inputting the plurality of sets of segmented images into the feature recognition model to output the parking features.

[0098] Further, the parking feature determination module 12 further comprises the following execution steps:

[0099] inputting the plurality of sets of segmented images into the feature recognition model;

[0100] outputting a parking state based on the dynamic feature recognition channel, wherein the parking state comprises a driving trajectory.

[0101] Based on the static feature recognition model, output feature state, the feature state includes violation features and vehicle basic information;

[0102] Based on the parking state and the feature state, determine the parking feature.

[0103] Further, the parking data chain module 13 includes the following execution steps:

[0104] Based on the driving trajectory, determine the parking time zone, the parking time zone is the time interval of entering and exiting the monitoring device parking space coverage domain;

[0105] Time sequence node segmentation is performed on the parking time zone to determine the link time sequence node, wherein the time interval of adjacent link time sequence nodes is not consistent;

[0106] Iterate through the link time sequence node, match and node identify the parking feature, and generate the parking data chain.

[0107] Further, the visual parking management module 15 includes the following execution steps:

[0108] Based on the parking data chain, generate pre-warning information for the violation feature to alert the vehicle violation;

[0109] If the violation feature meets the feature threshold, perform background record processing;

[0110] Based on the regional parking live map, perform visual display guidance of the terminal display device at the parking site.

[0111] Further, the present application embodiment also includes a monitoring device operation and maintenance module, which includes the following execution steps:

[0112] Set a directional operation and maintenance period to perform operation and maintenance management of the monitoring device;

[0113] Call the abnormal supervision records in the central management system within the predetermined time interval and perform frequency statistics to determine the abnormal record frequency, and if the abnormal record frequency is greater than or equal to the frequency threshold, generate a temporary operation and maintenance node;

[0114] Add the temporary operation and maintenance node to the directional operation and maintenance period.

[0115] Any step of the above method can be stored as computer instructions or programs in an unlimited computer memory and can be called and recognized by an unlimited computer processor to implement any method in the present application embodiment, without redundant limitations.

[0116] Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the application and its equivalents, the application is intended to cover all such modifications and variations as may fall within the scope of the application and its equivalents.

Claims

1. A parking management method based on high-position video, characterized in that, The methods include: Video is collected from the monitored area to determine the parking space monitoring video. The monitored area is equipped with at least one monitoring device, and each monitoring device corresponds to a parking space coverage area. The parking space monitoring video is fed back to the background processor, and the parking characteristics are determined frame by frame. The parking characteristics are marked with a time sequence. Based on the parking characteristics, a parking data chain is generated, and the parking data chain is identified as having violation information; Based on the parking space monitoring video, an area distribution map is determined and vacant parking spaces are marked, generating a real-time map of the area's parking spaces. This real-time map of the area's parking spaces is updated in real time. Based on the real-time map of parking spaces in the area and the parking data chain, the parking management is visualized and managed using the central management system. The frame-by-frame analysis to determine parking features includes the following methods: The parking space monitoring video is segmented frame by frame to determine multiple video frames; Based on the multiple video frames, edge detection is performed to extract the start frame image, the end frame image, and the key frame image. The image edge displacement values ​​of two adjacent frames in the key frame image meet the threshold standard. Image segmentation and feature recognition are performed on the start frame image, the end frame image, and the key frame image to determine parking features; The method for performing image segmentation and feature recognition on the start frame image, the end frame image, and the key frame image to determine parking features includes: The identification target is determined, and the starting frame image, the ending frame image and the key frame image are segmented based on the identification target to determine multiple sets of segmented images, wherein each set of segmented images corresponds to one identification target; Build a feature recognition model, including parallel dynamic feature recognition channels and static feature recognition channels; The multiple sets of segmented images are input into the feature recognition model, and the parking features are output. The methods include: Set a targeted maintenance cycle to perform maintenance management of the monitoring equipment; The abnormal monitoring records within a predetermined time interval of the central management system are called and frequency statistics are performed to determine the frequency of abnormal records. If the frequency of abnormal records is greater than or equal to the frequency threshold, a temporary operation and maintenance node is generated. Add the temporary maintenance node to the targeted maintenance cycle.

2. The method as described in claim 1, characterized in that, The method of inputting the multiple sets of segmented images into the feature recognition model and outputting the parking features includes: The multiple sets of segmented images are input into the feature recognition model; Based on the dynamic feature recognition channel, the parking status is output, including the driving trajectory; Based on the static feature recognition model, a feature status is output, which includes violation features and basic vehicle information. The parking feature is determined based on the parking state and the feature state.

3. The method as described in claim 2, characterized in that, The method for generating a parking data chain based on the parking features includes: Based on the driving trajectory, the parking time zone is determined, which is the time interval between entering and exiting the parking space coverage area of ​​the monitoring equipment; The parking time zone is segmented into time sequence nodes to determine the link time sequence nodes, wherein the time intervals of adjacent link time sequence nodes are not consistent; The parking data chain is generated by traversing the time-series nodes of the link, matching the parking features and identifying the nodes.

4. The method as described in claim 1, characterized in that, The methods include: Based on the parking data chain, early warning information is generated to alert vehicles to violations based on violation characteristics; If the violation characteristics meet the characteristic threshold, a background filing process will be initiated. Based on the actual parking space map of the area, a visual display guide is provided on the parking site using a terminal display device.

5. A parking management system based on high-position video, characterized in that, For implementing the parking management method based on high-view video as described in any one of claims 1 to 4, the system comprises: The monitoring video determination module is used to collect video of the monitored area and determine the monitoring video of the parking space. The monitored area is equipped with at least one monitoring device, and each monitoring device corresponds to a parking space coverage area. A parking feature determination module is used to feed back the parking space monitoring video to the background processor, analyze and determine parking features frame by frame, and the parking features are marked with a time sequence. A parking data chain module generates a parking data chain based on the parking characteristics, and the parking data chain identifies violation information. The parking space status map module determines the regional distribution map and marks vacant parking spaces based on the parking space monitoring video, and generates a regional parking space status map, which is updated in real time. A visual parking management module is used to perform visual display and parking management based on the real-time map of parking spaces in the area and the parking data chain, using the central management system.

Images