Inspection vehicle detects abnormal parking method and device, storage medium and electronic equipment

Abstract

The application relates to a method and device for detecting abnormal parking of an inspection vehicle, a storage medium and electronic equipment, wherein the method comprises the following steps: extracting parking space information of each parking space, chassis information and license plate information of each vehicle in a current frame picture from parking space video data; updating parking space tracking information of a corresponding parking space according to the parking space information to obtain target parking space tracking information, updating license plate tracking information of a corresponding vehicle according to the license plate information to obtain target license plate tracking information, and updating chassis tracking information of the corresponding vehicle according to the chassis information to obtain target chassis tracking information; matching target license plate tracking information corresponding to each target chassis tracking information and corresponding target parking space tracking information to finally determine matched chassis information; and determining whether the parking state of a vehicle corresponding to the latest license plate information is abnormal according to the relative position relationship between the matched chassis information and the latest parking space information. The application has the effect of covering more parking spaces for detecting the parking state.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of parking inspection, and particularly relates to a parking inspection vehicle abnormal parking detection method and device, a storage medium and an electronic device. BACKGROUND

[0002] At present, vehicle parking management is a typical problem in urban property management. The phenomenon of vehicle parking occupying the road and parking irregularly is common, which easily leads to low berth utilization rate. To solve this problem, staff need to continuously patrol and supervise for a long time, which is inefficient and consumes a lot of human resources. Therefore, how to better detect the vehicle parking state has become a problem to be solved.

[0003] The prior art mainly adopts a fixed camera to shoot image information or video information of berths in a fixed area of a parking lot, analyzes and processes the obtained image information or video information, and determines whether the parking state of the berths in the fixed area is abnormal. This method can only detect the berths in the fixed area, resulting in fewer berths covered by the detection of the parking state. SUMMARY

[0004] In order to cover more berths in the detection of the parking state, the present application provides a parking inspection vehicle abnormal parking detection method and device, a storage medium and an electronic device.

[0005] In a first aspect of the present application, a parking inspection vehicle abnormal parking detection method is provided, which specifically comprises:

[0006] acquiring berth video data through the camera, extracting berth information of each berth in a current frame picture from the berth video data, and extracting chassis information and license plate information of each vehicle in the current frame picture;

[0007] updating berth tracking information of the corresponding berth according to each berth information to obtain target berth tracking information of the corresponding berth, updating license plate tracking information of the corresponding vehicle according to each license plate information to obtain target license plate tracking information of the corresponding vehicle, and updating chassis tracking information of the corresponding vehicle according to each chassis information to obtain target chassis tracking information of the corresponding vehicle;

[0008] matching the latest license plate information in the target license plate tracking information corresponding to the latest chassis information in each target chassis tracking information to obtain matched chassis information, and determining the latest berth information in the target berth tracking information corresponding to the matched chassis information;

[0009] determining whether the parking state of the vehicle corresponding to the latest license plate information is abnormal according to the relative positional relationship between the matched chassis information and the corresponding latest berth information.

[0010] By adopting the technical solution, each parking space information is updated into the parking space tracking information of the corresponding parking space to obtain target parking space tracking information, each chassis information is updated into the chassis tracking information of the corresponding vehicle to obtain target chassis tracking information, and each license plate information is updated into the license plate tracking information of the corresponding vehicle to obtain target license plate tracking information, so that real-time tracking of the parking space, chassis and license plate appearing in the parking space video data shot by the inspection vehicle is realized. Then, the target license plate tracking information corresponding to each target chassis tracking information and the corresponding target parking space tracking information are matched, that is, the license plate and the vehicle are associated, and the vehicle and the parked parking space are determined. Finally, according to the relative position relationship between the matched chassis information and the corresponding latest parking space information, that is, the relative position relationship between the vehicle and the corresponding parking space, it is determined whether the parking state of the corresponding vehicle is abnormal, so that more parking spaces covered by the parking state are detected by the inspection vehicle.

[0011] Optionally, the extracting the chassis information and the license plate information of each vehicle in the current frame image specifically comprises:

[0012] The current frame image is input into the trained Deep3DBox algorithm to obtain the 3D bounding box of each vehicle in the current frame image.

[0013] The bottom of each 3D bounding box is projected onto a 2D plane to obtain the chassis information corresponding to each vehicle.

[0014] The current frame image is input into the trained license plate detection model to obtain the license plate information of each vehicle.

[0015] By adopting the technical solution, the trained Deep3DBox algorithm extracts the 3D bounding box of the vehicle appearing in the current frame image, that is, the frame of the vehicle, and then projects the bottom of each 3D bounding box onto a 2D plane, so that the chassis information of each vehicle, that is, the information of the vehicle chassis in the 2D plane, is obtained more quickly and accurately. Finally, the trained license plate detection model is used to extract the license plate information of the vehicle appearing in the current frame image.

[0016] Optionally, the updating the parking space tracking information of the corresponding parking space according to each parking space information to obtain target parking space tracking information of the corresponding parking space specifically comprises:

[0017] The first parking space quadrilateral corresponding to the latest parking space information in the parking space tracking information of each parking space is subjected to quadrilateral IOU calculation with the second parking space quadrilateral corresponding to each parking space information to obtain an IOU value.

[0018] Each IOU value is compared with an IOU threshold value to screen a parking space information set with an IOU value greater than the IOU threshold value.

[0019] Select the target parking space information with the maximum IOU value in the parking space information set to update the parking space tracking information of the corresponding parking space, and obtain the target parking space tracking information of the corresponding parking space.

[0020] By adopting the technical scheme, the first parking space quadrilateral and the second parking space quadrilateral corresponding to each parking space information in the current frame picture are subjected to IOU calculation to obtain an IOU value, so as to determine the parking space tracking information corresponding to each parking space information through the IOU value. If the IOU value is greater than an IOU threshold value, it is indicated that the corresponding parking space information and the current parking space tracking information are the same parking space with a high probability. Then, the target parking space information with the maximum IOU value in the parking space information set is selected, and the target parking space information and the current parking space tracking information are determined as the same parking space. The target parking space tracking information is obtained by updating the parking space tracking information of the corresponding parking space, so as to realize real-time tracking of the parking space.

[0021] Optionally, the license plate information includes a license plate number and license plate position information. The license plate tracking information of the corresponding vehicle is updated according to each license plate information, and target license plate tracking information of the corresponding vehicle is obtained. Specifically, the method comprises the following steps.

[0022] The license plate number with the highest frequency of occurrence in the license plate tracking information of each vehicle is determined as a voting license plate number.

[0023] Each license plate number is subjected to character bit-by-bit matching with the voting license plate number, and the number of matching characters is obtained.

[0024] According to the relative positional relationship between each license plate position information and the latest license plate position information in the license plate tracking information corresponding to the voting license plate number, the corresponding number of matching characters is adjusted, and a target number of matching characters is obtained.

[0025] In the license plate information with the target number of matching characters greater than a character number threshold value, the license plate information with the maximum target number of matching characters is selected to update the license plate tracking information of the corresponding vehicle, and target license plate tracking information of the corresponding vehicle is obtained.

[0026] By adopting the technical scheme, the license plate number in each detected license plate information is matched with the voting license plate number character by character, that is, the characters in the same position are compared to obtain the number of matched characters, that is, the number of characters that are the same. Then, according to the relative positional relationship between the license plate position information in each detected license plate information and the latest license plate position information in the currently compared license plate tracking information, each matched character number is optimized and adjusted to obtain a matched character number, so that it is more accurate and reasonable. If the target matched character number is greater than the character number threshold, it indicates that the license plate number is more similar to the voting license plate number. Finally, the license plate number with the maximum target matched character number is selected from the license plate numbers with the matched character number greater than the character number threshold, and it is determined that the license plate number is the same as the current license plate tracking information. The corresponding license plate information is updated to the current license plate tracking information, and real-time tracking of the license plate is realized.

[0027] Optionally, the target matched character number is obtained by adjusting the corresponding matched character number according to the relative positional relationship between the license plate position information and the latest license plate position information corresponding to the voting license plate number, and the adjusting specifically includes:

[0028] If the relative positional relationship is intersection, the corresponding matched character number is increased by a preset value to obtain the target matched character number.

[0029] If the relative positional relationship is no intersection, the corresponding matched character number is determined as the target matched character number.

[0030] By adopting the technical scheme, if the relative positional relationship between the license plate position information and the latest license plate position information is intersection, it indicates that the possibility of the license plate detected by the inspection vehicle being the same as the license plate corresponding to the current license plate tracking information is increased, and then the matched character number of the corresponding license plate number is increased to obtain the target matched character number. If the relative positional relationship is no intersection, the corresponding matched character number is maintained unchanged, that is, the matched character number is directly determined as the target matched character number, so that the license plate information that is the same as the current license plate tracking information is more accurately determined.

[0031] Optionally, the latest parking space information corresponds to a latest parking space, and the relative positional relationship between the matched chassis information and the corresponding latest parking space information is used to determine whether the parking state of the vehicle corresponding to the latest license plate information is abnormal, and the determining specifically includes:

[0032] If the relative positional relationship is within the corresponding latest parking space, the first parking state is determined to be normal, and a first cumulative number of the first parking state is counted.

[0033] If the relative positional relationship is intersected with the corresponding latest parking space, the second parking state is determined to be abnormal, and a second cumulative number of the second parking state is counted.

[0034] If the relative position relationship is not in the corresponding latest parking position, a third parking state is determined to be abnormal, and a third cumulative number of the third parking state is counted;

[0035] A continuous frame number of the target license plate tracking information corresponding to the matched chassis information not being updated is counted, and if the continuous frame number is greater than a preset frame number, a parking state corresponding to a maximum value of the first cumulative number, the second cumulative number and the third cumulative number is determined as a parking state of a vehicle corresponding to the latest license plate information.

[0036] By using the above technical solution, the relationship between the vehicle corresponding to the matched chassis information and the parking position is determined in real time from the current frame picture, the vehicle chassis is in the latest parking position, which indicates the parking state, the first parking state is determined to be normal, the vehicle chassis intersects with the latest parking position, which indicates that the vehicle may press the parking position line or park across the parking position, the second parking state is determined to be abnormal, and the vehicle chassis is not in the latest parking position, which indicates that the vehicle has not been parked in the parking position, and the third parking state at this time is determined to be abnormal. Then, if the continuous frame number is greater than the preset frame number, it indicates that the vehicle corresponding to the matched chassis information has not appeared in the video picture for a long time, and then it is indicated that the inspection vehicle has left, the parking state of the corresponding vehicle is finally determined at this time, the parking state corresponding to the maximum value of the first cumulative number, the second cumulative number and the third cumulative number is selected, that is, the state of the corresponding vehicle is most likely, so that the parking state of the vehicle in more parking positions can be determined more accurately.

[0037] Optionally, the latest chassis information includes a chassis quadrilateral, the latest license plate information in the target license plate tracking information corresponding to the latest chassis information in each target chassis tracking information is matched to obtain matched chassis information, and the matching specifically includes:

[0038] Two vehicle head points corresponding to the latest chassis information are determined.

[0039] The two vehicle head points are translated upward along a vertical direction of a plane where the chassis quadrilateral is located to obtain a vehicle head plane.

[0040] It is judged whether the license plate corresponding to each latest license plate information is in the vehicle head plane, if yes, the corresponding latest license plate information and the latest chassis information are matched to obtain the matched chassis information.

[0041] By adopting the technical scheme, after two head points corresponding to the latest chassis information, i.e., two head points of the chassis quadrilateral are determined, the two head points are translated upward to obtain a head plane where the license plate is located, if the target license plate tracking information corresponds to the license plate located in the head plane, it is indicated that the target license plate tracking information and the current target chassis tracking information correspond to the same vehicle, and the two are in a matching relationship. Thus, real-time tracking of the vehicle is realized.

[0042] In a second aspect of the present application, an abnormal parking detection device for a patrol vehicle is provided, and specifically includes:

[0043] An information acquisition module is configured to acquire parking space video data through the camera, extract parking space information of each parking space in a current frame from the parking space video data, and extract chassis information and license plate information of each vehicle in the current frame;

[0044] A tracking update module is configured to update parking space tracking information of a corresponding parking space according to each parking space information to obtain target parking space tracking information of the corresponding parking space, update license plate tracking information of a corresponding vehicle according to each license plate information to obtain target license plate tracking information of the corresponding vehicle, and update chassis tracking information of the corresponding vehicle according to each chassis information to obtain target chassis tracking information of the corresponding vehicle;

[0045] An information matching module is configured to match latest license plate information in target license plate tracking information corresponding to latest chassis information in each target chassis tracking information to obtain matched chassis information, and determine latest parking space information in target parking space tracking information corresponding to the matched chassis information;

[0046] An abnormality judgment module is configured to determine whether a parking state of a vehicle corresponding to the latest license plate information is abnormal according to a relative position relationship between the matched chassis information and corresponding latest parking space information.

[0047] By adopting the technical scheme, after the information acquisition module acquires the parking space information, the license plate information and the chassis information, the tracking update module updates the parking space tracking information of the corresponding parking space, the license plate tracking information of the corresponding vehicle and the chassis tracking information of the corresponding vehicle, then the information matching module matches, for each target chassis tracking information, corresponding target license plate tracking information and corresponding target parking space tracking information. Finally, the abnormality judgment module determines whether the parking state of the vehicle corresponding to the latest license plate information is abnormal according to the relative position relationship between the matched chassis information and corresponding latest parking space information.

[0048] In summary, the present application has at least one of the following beneficial technical effects:

[0049] The system updates the parking space information for each parking space to its corresponding parking space tracking information, obtaining the target parking space tracking information. Similarly, it updates the chassis information for each vehicle to its corresponding chassis tracking information, obtaining the target chassis tracking information. Finally, it updates the license plate information for each vehicle to its corresponding license plate tracking information, obtaining the target license plate tracking information. This allows for real-time tracking of parking spaces, chassis, and license plates captured by the inspection vehicle in video data. Next, it matches the target license plate tracking information with the target parking space tracking information for each target chassis tracking information, establishing a correspondence between the license plate and the vehicle, and then determining the vehicle and its parking space. Finally, based on the relative positional relationship between the matched chassis information and the latest parking space information (i.e., the relative positional relationship between the vehicle and its corresponding parking space), it determines whether the vehicle's parking status is abnormal. This allows the inspection vehicle to detect a wider range of parking spaces. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the architecture of an inspection vehicle detection system for abnormal parking provided in an embodiment of this application;

[0051] Figure 2 This is a flowchart illustrating a method for detecting abnormal parking by an inspection vehicle, as provided in an embodiment of this application.

[0052] Figure 3 This is a flowchart illustrating a patrol vehicle detecting abnormal parking, as provided in an embodiment of this application.

[0053] Figure 4 This is a schematic diagram illustrating different tracking information updates and matching methods provided in an embodiment of this application;

[0054] Figure 5 This is a flowchart illustrating another method for detecting abnormal parking by an inspection vehicle provided in an embodiment of this application;

[0055] Figure 6 This is a schematic diagram illustrating another different tracking information update and matching method provided in an embodiment of this application;

[0056] Figure 7 This is a schematic diagram of the structure of an inspection vehicle abnormal parking detection device provided in an embodiment of this application.

[0057] Explanation of reference numerals in the attached diagram: 11. Information acquisition module; 12. Tracking and updating module; 13. Information matching module; 14. Anomaly detection module. Detailed Implementation

[0058] In order for those skilled in the art to better understand the technical solutions in the specification, the technical solutions in the specification will be clearly and completely described below in conjunction with the drawings in the specification. Obviously, the described embodiments are only some of the embodiments of the present application, not all.

[0059] In the description of the embodiments of the present application, the words such as "exemplary", "for example", or "for instance" are used to mean that the examples, instances, or illustrations are examples, instances, or illustrations. Any embodiment or design solution described as "exemplary", "for example", or "for instance" in the embodiments of the present application should not be interpreted as being more preferred or having more advantages than other embodiments or design solutions. Rather, the words such as "exemplary", "for example", or "for instance" are intended to present the relevant concept in a specific manner.

[0060] Referring to Figure 1 The embodiments of the present application disclose an architecture schematic diagram of a patrol vehicle detection abnormal parking system, comprising a patrol vehicle and a terminal. The patrol vehicle detection abnormal parking method is applied to the patrol vehicle. The terminal can be an electronic device such as a mobile phone, a tablet computer, an electronic book reader, a multimedia playing device, a wearable device, a personal computer (PC), etc. The patrol vehicle can be an AI video patrol vehicle, and a camera is arranged on the patrol vehicle. The terminal and the patrol vehicle are directly or indirectly connected through a wireless network.

[0061] Specifically, when the patrol vehicle moves for inspection in a park or a parking lot, the camera captures real-time videos of parking spaces in different areas, obtains video data, then extracts parking space information of each parking space, chassis information and license plate information of each vehicle in the current frame of the video data in real time, then matches the parking space information, the chassis information and the license plate information, and finally determines whether the parking state of the corresponding vehicle on the parking space is abnormal according to the parking space information and the corresponding chassis information, and sends the parking state to the terminal in real time to notify the personnel to timely handle the vehicle with abnormal parking, so as to avoid low utilization rate of parking spaces, and at the same time, the range covered when detecting the parking state is wider and the parking spaces involved are more.

[0062] Referring to Figure 2 The embodiments of the present application disclose a flowchart of a patrol vehicle detection abnormal parking method, which can be realized by relying on a computer program, and can also run on a patrol vehicle detection abnormal parking device based on the von Neumann system. The computer program can be integrated in an application, or can run as an independent tool class application. Specifically, the computer program comprises the following steps:

[0063] S101: acquiring parking space video data through the camera, extracting parking space information of each parking space in the current frame from the parking space video data, and extracting chassis information and license plate information of each vehicle in the current frame.

[0064] In an implementable embodiment, the extracting chassis information and license plate information of each vehicle in the current frame picture comprises: inputting the current frame picture into the trained Deep3DBox algorithm to obtain a 3D bounding box of each vehicle in the current frame picture;

[0065] projecting the bottom of each 3D bounding box to a 2D plane to obtain chassis information corresponding to each vehicle;

[0066] inputting the current frame picture into the trained license plate detection model to obtain license plate information of each vehicle.

[0067] Specifically, the parking space video data is obtained in real time by a camera. The parking space video data is real-time video of a region with parking spaces in a park or parking lot shot by a patrol vehicle. The picture corresponding to the parking space video data may only have parking spaces, or may have parking spaces, parked vehicles and vehicles being parked, etc.

[0068] Then, the parking space information of each parking space in the current frame picture of the parking space video data, i.e., the frame picture corresponding to the current time, is extracted. In the application embodiment, the parking space information includes four inner corner points and a 2D position of the parking space. One feasible way to extract the parking space information of each parking space is to extract it from the current frame picture by using a trained parking space extraction model. The parking space extraction model adopts a convolutional neural network model or a bp neural network model. The specific process of training the parking space extraction model is as follows: 1000 groups of parking space images in different scenarios such as on-street parking, closed parking lot, semi-closed parking lot, different states such as vehicle in parking space, no vehicle in parking space, and foreign matter occupying, and different weather conditions are collected, the 2D position of the parking space and the inner corner points of the parking space in the parking space images are labeled to obtain labeled images, and then the labeled images are used as training data to train the parking space extraction model by using the yolo target detection algorithm until the model converges to obtain the trained parking space extraction model.

[0069] The Deep3DBox algorithm is a three-dimensional target detection algorithm, which is mainly used to predict the boundary box and attitude angle of a three-dimensional target. The current frame picture is input into the trained Deep3DBox algorithm, and a 3D bounding box of each vehicle in the current frame picture is extracted therefrom. The 3D bounding box is a frame body surrounding the vehicle, which is represented by a cube or other geometric body. Then, the 3D bounding box of each vehicle is projected to a 2D plane. One feasible way is to calculate a projection matrix by using an orthogonal projection algorithm. The projection matrix is a transformation matrix for projecting a 3D point from a 3D space to a 2D image space. Then, each 3D point in the 3D bounding box is multiplied by the projection matrix to obtain the corresponding point on the 2D plane, and thus the chassis information corresponding to each vehicle, i.e., the 2D position of the chassis of each vehicle on the 2D plane, is obtained.

[0070] In addition, the current frame picture is input again into the trained license plate detection model, so as to extract license plate information of each vehicle from the current frame picture, the license plate information including vehicle number and license plate position information. The license plate detection model adopts a convolutional neural network model or a recurrent neural network model. The process of training the license plate detection model is as follows: based on a yolo target detection algorithm and an ORC character recognition algorithm, image samples labeled with license plate numbers and license plate positions are used to train the license plate detection model, and finally the trained license plate detection model is obtained.

[0071] S102: According to each berth information, update the berth tracking information of the corresponding berth to obtain the target berth tracking information of the corresponding berth, update the license plate tracking information of the corresponding vehicle according to each license plate information to obtain the target license plate tracking information of the corresponding vehicle, and update the chassis tracking information of the corresponding vehicle according to each chassis information to obtain the target chassis tracking information of the corresponding vehicle.

[0072] In an implementable embodiment, the first berth quadrilateral corresponding to the latest berth information in the berth tracking information of each berth is subjected to quadrilateral IOU calculation with the second berth quadrilateral corresponding to each berth information, to obtain an IOU value;

[0073] Each IOU value is compared with an IOU threshold value, and a berth information set with an IOU value greater than the IOU threshold value is screened out;

[0074] The target berth information with the maximum IOU value in the berth information set is selected and updated to the berth tracking information of the corresponding berth, to obtain the target berth tracking information of the corresponding berth.

[0075] Specifically, each berth corresponds to a berth tracking information, and the berth tracking information includes berth information on multiple historical trajectories, i.e., a berth quadrilateral formed by four inner corner points of the berth. In the application embodiment, the latest berth information is determined as follows: according to the pixel displacement of the berth quadrilateral of the historical trajectory and the displacement time, the moving speed of each vertex of the berth quadrilateral can be calculated. According to the moving speed and the time difference between the current frame and the previous frame, the possible positions of each vertex of the berth quadrilateral in the current frame can be calculated, and finally the latest berth information is obtained.

[0076] After obtaining the berth information of multiple berths in the current frame picture, the first berth quadrilateral corresponding to the latest berth information in the berth tracking information of each berth established in advance is subjected to quadrilateral IOU calculation with the second berth quadrilateral corresponding to the obtained each berth information, to obtain multiple IOU values. The quadrilateral IOU calculation is specifically: the intersection of the first berth quadrilateral and the second berth quadrilateral is divided by the union of the first berth quadrilateral and the second berth quadrilateral. The quadrilateral IOU is an index for measuring the overlapping degree of two irregular quadrilaterals. The greater the IOU value is, the higher the overlapping degree of the first berth quadrilateral and the second berth quadrilateral is.

[0077] Then, a set of berth information with an IOU value greater than an IOU threshold value is selected from the berth information. The IOU threshold value is a critical value for judging whether the berth corresponding to the second berth quadrilateral is similar to the berth corresponding to the first berth quadrilateral. If the IOU value is greater than the IOU threshold value, it indicates that the corresponding berth information is more similar to the latest berth information in the current berth tracking information. Finally, the target berth information with the greatest IOU value is selected from the set of berth information, and it is determined that the berth corresponding to the target berth information and the berth corresponding to the current berth tracking information are the same berth. Then, the berth information with the greatest IOU value is updated to the berth tracking information of the corresponding berth, to obtain the target berth tracking information.

[0078] It should be noted that if there is a berth information in the obtained berth information from the current frame picture, which is not the same as the berth corresponding to each established berth tracking information, a new berth tracking information corresponding to the berth information is re-established.

[0079] After the target berth tracking information corresponding to the berth is determined, the license plate tracking information corresponding to the vehicle needs to be updated. The license plate tracking information includes each historical license plate information corresponding to the historical trajectory. The historical license plate number with the highest frequency of occurrence in the license plate tracking information is counted, and the characters of the historical license plate number and the license plate number in each obtained license plate information are compared bit by bit. The license plate information corresponding to the license plate number with the most consistent characters is updated to the license plate tracking information, i.e., the license plate tracking information corresponding to the vehicle is obtained. The most consistent characters indicate that the license plate number and the historical license plate number are the same license plate.

[0080] If there is a license plate information in the obtained license plate information from the current frame picture, which is not the same as the license plate number corresponding to each established license plate tracking information, a new license plate tracking information corresponding to the license plate information is re-established.

[0081] Finally, the chassis tracking information for the corresponding vehicle is updated. For details, please refer to the process of determining the target parking space tracking information. This involves performing a quadrilateral Interchange of Union (IOU) calculation between the acquired chassis information and the latest chassis information in each established chassis tracking information set. Each chassis information set and the latest chassis information set are essentially quadrilaterals corresponding to the chassis. Finally, the chassis information set with the highest IOU value (i.e., the same chassis as the latest chassis information) is selected and updated in this chassis tracking information set, thus obtaining the target chassis tracking information for the corresponding vehicle.

[0082] S103: Match the latest license plate information in the target license plate tracking information corresponding to the latest chassis information in each target chassis tracking information to obtain the matched chassis information, and determine the latest parking space information in the target parking space tracking information corresponding to the matched chassis information.

[0083] In one feasible implementation, the latest chassis information includes the latest chassis and two front points, and the two front points corresponding to the latest chassis information are determined.

[0084] The two front points are translated upwards along the vertical direction of the plane where the latest chassis is located to obtain the front plane;

[0085] Determine whether the license plate corresponding to each latest license plate information is on the front plane of the vehicle. If it is, establish a matching relationship between the latest license plate information and the latest chassis information to obtain the matched chassis information.

[0086] Specifically, such as Figure 3 As shown, once the target chassis tracking information, target license plate tracking information, and target parking space tracking information are determined, it indicates that the license plate tracking information, chassis tracking information, and parking space tracking information corresponding to the parking spaces and vehicles in the current frame have been updated in real time. Next, it is necessary to establish matching relationships between each target chassis tracking information, each target license plate tracking information, and each target parking space tracking information; that is, to match the license plates and chassis in the current frame, and then match the chassis with the matched license plates with the corresponding parking spaces.

[0087] First, the latest chassis information is selected from the tracking information of each target chassis. Then, the license plate information corresponding to this latest chassis information is matched from the latest license plate information in the tracking information of multiple targets; that is, the chassis and license plate information of the same vehicle are bound together. One feasible matching method is:

[0088] Since the latest chassis information is essentially a chassis quadrangle corresponding to a 2D chassis (formed by a chassis contour), the chassis quadrangle includes four points, divided into two vehicle head points and two vehicle position points, therefore, two points on the shorter side of the chassis quadrangle and closest to the bottom of the current frame are selected as the two vehicle head points, and in other embodiments, the two points can also be used as the two vehicle position points, because when a vehicle is parked, the vehicle head faces inwards and the vehicle position faces outwards.

[0089] The two vehicle head points are simultaneously translated upwards by a preset distance along the vertical direction of the plane of the chassis quadrangle, and the two vehicle head points after translation and the initial two vehicle head points form a new plane, i.e., a vehicle head plane, which is perpendicular to the plane of the chassis quadrangle.

[0090] Then, according to the license plate position information corresponding to the latest license plate information in each target license plate tracking information, it is determined whether the corresponding license plate is located on the vehicle head plane, if it is located on the vehicle head plane, it means that the license plate and the chassis belong to the same vehicle, therefore, the latest license plate information and the latest chassis information are matched to obtain matched chassis information, and the matching of the next target chassis tracking information is continued. If it is not located on the vehicle head plane, the next latest license plate information in the target license plate tracking information is selected for matching.

[0091] Finally, the matched chassis information is determined with the latest parking space information in the corresponding target parking space tracking information, and a feasible matching method is to calculate the distance between the vehicle corresponding to the matched chassis information and the parking space corresponding to the latest parking space information in each target parking space tracking information, and the latest parking space information with the smallest distance is selected to establish a matching relationship with the matched chassis information, i.e., the vehicle corresponding to the matched chassis information is entering or parked in the parking space.

[0092] For example, there are vehicle 1, vehicle 2, parking space 1, parking space 2 and parking space 3 in the current frame, and license plate information 1, license plate information 2, chassis information 1, chassis information 2, parking space information 1, parking space information 2 and parking space information 3 are extracted, and the license plate information is updated to the corresponding license plate tracking information, the chassis information is updated to the corresponding chassis tracking information, and the parking space information is updated to the corresponding parking space tracking information, then the license plate is matched with the chassis, and the matched chassis is matched with the corresponding parking space, for details, please refer to Figure 4 .

[0093] S104: According to the relative position relationship between the matched chassis information and the corresponding latest parking space information, it is determined whether the parking state of the vehicle corresponding to the latest license plate information is abnormal.

[0094] Specifically, the relative position relationship of the matched chassis information and the corresponding latest parking space information, i.e., the relative position relationship of the matched chassis quadrangle corresponding to the matched chassis information and the latest parking space quadrangle corresponding to the latest parking space information, if the matched chassis quadrangle is located in the latest parking space quadrangle or the matched chassis quadrangle is completely contained in the latest parking space quadrangle, it indicates that the chassis is completely located in the parking space, and the vehicle is normally parked in the parking space, which belongs to normal parking. If the matched chassis quadrangle and the latest parking space quadrangle do not intersect, it indicates that the vehicle is located outside the parking space, and is determined as abnormal parking. If the matched chassis quadrangle and the latest parking space quadrangle have a partial intersection, it indicates that the vehicle is not completely parked in the parking space, and is determined as abnormal parking.

[0095] Referring to Figure 5 The embodiment of the present application discloses another flowchart of the abnormal parking detection method of the inspection vehicle, which can be realized by relying on a computer program and can also run on an inspection vehicle abnormal parking detection device based on the von Neumann system. The computer program can be integrated in an application or can run as an independent tool application. Specifically, the computer program comprises the following steps:

[0096] S201: acquiring parking space video data through a camera, extracting parking space information of each parking space in a current frame picture from the parking space video data, and extracting chassis information and license plate information of each vehicle in the current frame picture.

[0097] S202: updating parking space tracking information of a corresponding parking space according to each parking space information to obtain target parking space tracking information of the corresponding parking space.

[0098] Specifically, reference can be made to steps S101-S102, which will not be repeated here.

[0099] S203: determining a voted license plate number in license plate tracking information of each vehicle, the voted license plate number being a license plate number with the highest appearance frequency.

[0100] S204: performing character bit-by-bit matching on each license plate number and the voted license plate number to obtain a matching character number.

[0101] Specifically, the license plate information of each vehicle is acquired from the current frame picture, which includes a license plate number and license plate position information, the license plate position information referring to spatial position information of the license plate. A license plate tracking information of a vehicle is sequentially selected from the established license plate tracking information of a plurality of vehicles, the appearance frequency of different license plate numbers in the license plate tracking information is counted, and the license plate number with the highest appearance frequency is determined as the voted license plate number, i.e., the correct license plate number of the corresponding vehicle. In other embodiments, the character with the highest appearance frequency at each position of the license plate can also be determined, and each character is matched with its position in the license plate to form the voted license plate number.

[0102] Then, the characters in each license plate number corresponding to the current frame picture are matched with the characters in the voting license plate number bit by bit, that is, the characters in the same position of the license plate are compared, and the number of matching characters in each position corresponding to the position of the voting license plate number is counted. The more the number of matching characters is, the more likely the corresponding license plate number corresponds to the same license plate as the voting license plate number. For example, the voting license plate number is E A0P099, and the license plate number 1 is E A0P09?. The number of matching characters of the two is 6, which are E, A, 0, P, 0, and 9.

[0103] S205: According to the relative position relationship between each license plate position information and the latest license plate position information in the license plate tracking information corresponding to the voting license plate number, the corresponding number of matching characters is adjusted to obtain a target number of matching characters.

[0104] In an implementable embodiment, if the relative position relationship is intersection, the corresponding number of matching characters is increased by a preset value to obtain the target number of matching characters.

[0105] If the relative position relationship is no intersection, the corresponding number of matching characters is determined as the target number of matching characters.

[0106] Specifically, the current frame picture corresponding to each license plate position information is compared with the latest license plate position information corresponding to the latest license plate information in the current license plate tracking information. If the relative position relationship is intersection, it is indicated that the license plates corresponding to the two have intersection, that is, they are the same license plate with high probability. Then, the number of matching characters of the license plate number corresponding to the license plate position information is increased by a preset value to obtain a target number of matching characters, that is, the probability that the license plate corresponding to the license plate number is the same license plate as the license plate corresponding to the current license plate tracking information is increased. In the embodiments of the present application, the preset value can be 1, and in other embodiments, it can also be 2 or 3.

[0107] If the relative position relationship is no intersection, it is indicated that the license plates corresponding to the two have no intersection, that is, they are not the same license plate with high probability. Then, the number of matching characters of the license plate number corresponding to the license plate position information is not adjusted, and is directly determined as the target number of matching characters.

[0108] S206: In the license plate information with the target number of matching characters greater than the threshold number of characters, the license plate information with the maximum target number of matching characters is selected and updated to the license plate tracking information corresponding to the vehicle to obtain target license plate tracking information of the corresponding vehicle.

[0109] Specifically, after the target matching character number corresponding to the license plate information of each vehicle is detected, the target matching character number is compared with the character number threshold. If the target matching character number is greater than the character number threshold, it indicates that the corresponding license plate number is similar to the voting license plate number. Then, the license plate information with the maximum target matching character number is selected from the license plate information with the target matching character number greater than the character number threshold, and is updated to the current license plate tracking information, that is, the corresponding license plates of the two are the same license plate. Finally, the target license plate tracking information of the corresponding vehicle is obtained.

[0110] If each target matching character number is not greater than the character number threshold, the next established license plate tracking information is continuously selected for updating processing.

[0111] S207: The chassis tracking information of the corresponding vehicle is updated according to each chassis information, and the target chassis tracking information of the corresponding vehicle is obtained.

[0112] Specifically, reference can be made to step S102, which will not be repeated here.

[0113] S208: The latest license plate information in the target license plate tracking information corresponding to the latest chassis information in each target chassis tracking information is matched, the matched chassis information is obtained, and the latest parking space information in the target parking space tracking information corresponding to the matched chassis information is determined.

[0114] Specifically, reference can be made to step S103, which will not be repeated here.

[0115] S209: Whether the parking state of the vehicle corresponding to the latest license plate information is abnormal is determined according to the relative positional relationship between the matched chassis information and the corresponding latest parking space information.

[0116] In an implementable embodiment, if the relative positional relationship is in the corresponding latest parking space, it is determined that the first parking state is normal, and the first cumulative number of the first parking state is counted.

[0117] If the relative positional relationship is intersected with the corresponding latest parking space, it is determined that the second parking state is abnormal, and the second cumulative number of the second parking state is counted.

[0118] If the relative positional relationship is not in the corresponding latest parking space, it is determined that the third parking state is abnormal, and the third cumulative number of the third parking state is counted.

[0119] The continuous frame number of the target license plate tracking information corresponding to the matched latest chassis information which is not updated is counted. If the continuous frame number is greater than the preset frame number, the parking state corresponding to the maximum value of the first cumulative number, the second cumulative number and the third cumulative number is determined as the parking state of the vehicle corresponding to the latest license plate information.

[0120] Specifically, as Figure 6As shown, since the inspection vehicle is shooting the parking space video in real time, the relative position relationship between the matched chassis information and the corresponding latest parking space information is also real-time changing, wherein the latest parking space information corresponds to the latest parking space, for example, in the shot parking space video, the first frame picture is that the vehicle A is located outside the parking space a, the second frame picture is that the vehicle A is approaching the parking space a, the third frame picture may be that the vehicle A is entering the parking space a, and the fourth frame picture may be that the vehicle A is completely in the parking space a. In view of this, instead of focusing on the current frame picture, each frame picture after the current frame picture is considered. If the chassis corresponding to the matched chassis information is located in the latest parking space, it is indicated that the first parking state is a normal state, and the first cumulative number of the first parking state is continuously counted, i.e. the number of times in each frame picture.

[0121] Starting from the current frame picture, the relationship between the chassis corresponding to the matched chassis information and the latest parking space is determined in real time. If the chassis corresponding to the matched chassis information intersects with the latest parking space, it is indicated that the corresponding vehicle may be parked in the line-pressing parking space or the cross-parking space. If the chassis intersects with another parking space in addition to the latest parking space, it is determined that the vehicle is parked in the cross-parking space, otherwise, it is determined that the vehicle is parked in the line-pressing parking space. Finally, the second parking state is determined to be an abnormal state, and the second cumulative number of the second parking state is continuously counted.

[0122] If the chassis corresponding to the matched chassis information does not intersect with the latest parking space or has no overlapping region, it is indicated that the vehicle is outside the latest parking space. The third parking state is determined to be an abnormal state, and the third cumulative number of the third parking state is continuously counted.

[0123] Finally, the continuous frame number of the target license plate tracking information corresponding to the matched chassis information which is not updated is counted. If the continuous frame number is greater than a preset frame number, it is indicated that the corresponding vehicle does not exist in the video picture shot by the camera in real time, which indicates that the inspection vehicle has moved to other areas, and further indicates that the vehicle has completed the parking action. Then, the determination of the final parking state is performed, i.e. the maximum value of the first cumulative number, the second cumulative number and the third cumulative number is selected, and the parking state corresponding to the maximum value is taken as the parking state of the corresponding vehicle.

[0124] The implementation principle of the method for detecting abnormal parking of the inspection vehicle in the embodiments of the present application is as follows: each parking space information is updated into the parking space tracking information of the corresponding parking space to obtain target parking space tracking information, each chassis information is updated into the chassis tracking information of the corresponding vehicle to obtain target chassis tracking information, and each license plate information is updated into the license plate tracking information of the corresponding vehicle to obtain target license plate tracking information, so that real-time tracking of the parking space, chassis and license plate appearing in the parking space video data shot by the inspection vehicle is realized. Then, the target license plate tracking information corresponding to each target chassis tracking information and the corresponding target parking space tracking information are matched, that is, the license plate and the vehicle are associated, and the vehicle and the parked parking space are determined. Finally, according to the relative position relationship between the matched chassis information and the corresponding latest parking space information, that is, the relative position relationship between the vehicle and the corresponding parking space, it is determined whether the parking state of the corresponding vehicle is abnormal, so that the parking state is detected by the inspection vehicle, and more parking spaces are covered.

[0125] The following is an apparatus embodiment of the present application, which can be used to execute the method embodiments of the present application. For details not disclosed in the apparatus embodiments of the present application, refer to the method embodiments of the present application.

[0126] Please refer to Figure 7 The structure diagram of the inspection vehicle detection abnormal parking device provided by the embodiments of the present application is shown. The inspection vehicle detection abnormal parking device can be realized by software, hardware or a combination of the two to become all or part of the device. The device 1 comprises an information acquisition module 11, a tracking update module 12, an information matching module 13 and an abnormality judgment module 14.

[0127] The information acquisition module 11 is configured to acquire parking space video data through a camera, extract parking space information of each parking space in the current frame picture from the parking space video data, and extract chassis information and license plate information of each vehicle in the current frame picture;

[0128] The tracking update module 12 is configured to update the parking space tracking information of the corresponding parking space according to the parking space information to obtain target parking space tracking information of the corresponding parking space, update the license plate tracking information of the corresponding vehicle according to the license plate information to obtain target license plate tracking information of the corresponding vehicle, and update the chassis tracking information of the corresponding vehicle according to the chassis information to obtain target chassis tracking information of the corresponding vehicle;

[0129] The information matching module 13 is configured to match the latest license plate information in the target license plate tracking information corresponding to the latest chassis information in each target chassis tracking information to obtain matched chassis information, and determine the latest parking space information in the target parking space tracking information corresponding to the matched chassis information;

[0130] The abnormality judgment module 14 is configured to determine whether the parking state of the vehicle corresponding to the latest license plate information is abnormal according to the relative position relationship between the matched chassis information and the corresponding latest parking space information.

[0131] Optionally, the information acquisition module 11 is specifically configured to:

[0132] input the current frame picture into the trained Deep3DBox algorithm to obtain the 3D bounding box of each vehicle in the current frame picture;

[0133] project the bottom of each 3D bounding box to a 2D plane to obtain the chassis information corresponding to each vehicle;

[0134] input the current frame picture into the trained license plate detection model to obtain the license plate information of each vehicle.

[0135] Optionally, the tracking updating module 12 is specifically configured to:

[0136] perform quadrilateral IOU calculation on the first parking space quadrilateral corresponding to the latest parking space information in the parking space tracking information of each parking space and the second parking space quadrilateral corresponding to each parking space information to obtain an IOU value;

[0137] compare each IOU value with an IOU threshold value to screen a parking space information set with an IOU value greater than the IOU threshold value;

[0138] select target parking space information with the maximum IOU value in the parking space information set to update to the parking space tracking information of the corresponding parking space to obtain target parking space tracking information of the corresponding parking space.

[0139] Optionally, the tracking updating module 12 is specifically configured to:

[0140] determine the voting license plate number in the license plate tracking information of each vehicle, the voting license plate number being the license plate number with the highest appearance frequency;

[0141] perform character bit-by-bit matching on each license plate number and the voting license plate number to obtain a matching character number;

[0142] adjust the corresponding matching character number according to the relative position relationship between the license plate position information and the latest license plate position information in the license plate tracking information corresponding to the voting license plate number to obtain a target matching character number;

[0143] select license plate information with the maximum target matching character number from the license plate information with the target matching character number greater than a character number threshold value to update to the license plate tracking information of the corresponding vehicle to obtain target license plate tracking information of the corresponding vehicle.

[0144] Optionally, the tracking updating module 12 is specifically configured to:

[0145] If the relative position relationship is intersection, the corresponding matching character number is increased by a preset value to obtain a target matching character number.

[0146] If the relative position relationship is no intersection, the corresponding matching character number is determined as the target matching character number.

[0147] Optionally, the anomaly judgment module 14 is specifically configured to:

[0148] If the relative position relationship is in the corresponding latest parking position, the first parking state is determined as normal, and a first cumulative number of the first parking state is counted.

[0149] If the relative position relationship is intersected with the corresponding latest parking position, the second parking state is determined as abnormal, and a second cumulative number of the second parking state is counted.

[0150] If the relative position relationship is not in the corresponding latest parking position, the third parking state is determined as abnormal, and a third cumulative number of the third parking state is counted.

[0151] The sustained frame number of the target license plate tracking information of the matched chassis information not being updated is counted, and if the sustained frame number is greater than a preset frame number, the parking state corresponding to the maximum value of the first cumulative number, the second cumulative number and the third cumulative number is determined as the parking state of the vehicle corresponding to the latest license plate information.

[0152] Optionally, the information matching module 13 is specifically configured to:

[0153] The two vehicle head points corresponding to the latest chassis information are determined.

[0154] The two vehicle head points are translated upward along the vertical direction of the plane where the chassis quadrilateral is located to obtain a vehicle head plane.

[0155] It is determined whether the license plate corresponding to each latest license plate information is in the vehicle head plane, and if yes, the corresponding latest license plate information and the latest chassis information are matched to obtain matched chassis information.

[0156] It should be noted that the above embodiment provides a patrol vehicle detection abnormal parking device, which is used to execute the patrol vehicle detection abnormal parking method. In the actual application, the above functions can be completed by different functional modules according to the needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the patrol vehicle detection abnormal parking device and the patrol vehicle detection abnormal parking method provided in the above embodiment belong to the same concept, and the implementation process is detailed in the method embodiment. Here, it is not repeated.

[0157] The embodiment of the present application further discloses a computer readable storage medium, and the computer readable storage medium stores a computer program.

[0158] The computer program can be stored in the computer readable medium, and the computer program includes computer program code, which can be in the form of source code, object code, executable file or some intermediate form of code, etc. The computer readable medium includes any entity or device, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier wave signal, telecommunication signal and software distribution medium, etc. that can carry the computer program code. It should be noted that the computer readable medium includes but is not limited to the above-mentioned components.

[0159] The computer readable storage medium stores the method for detecting abnormal parking of the inspection vehicle in the computer readable storage medium, and is loaded and executed on the processor to facilitate storage and application of the method.

[0160] The embodiment of the present application further discloses an electronic device, and the computer readable storage medium stores a computer program, and the computer program is loaded and executed on the processor to adopt the method for detecting abnormal parking of the inspection vehicle.

[0161] The electronic device can be a desktop computer, a notebook computer or a cloud server, and the electronic device includes but is not limited to a processor and a memory. For example, the electronic device can further include an input / output device, a network access device and a bus, etc.

[0162] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), ready programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The present application does not limit the processor.

[0163] The memory can be an internal storage unit of the electronic device, for example, a hard disk or a memory of the electronic device, or an external storage device of the electronic device, for example, a plug-in hard disk, a smart memory card (SMC), a secure digital card (SD), or a flash memory card (FC) equipped on the electronic device, or a combination of the internal storage unit and the external storage device. The memory is configured to store a computer program and other programs and data required by the electronic device, and can be used to temporarily store data that has been output or will be output. The present application is not limited in this regard.

[0164] The method for detecting an abnormal stop of the inspection vehicle is stored in the memory of the electronic device and loaded and executed on the processor of the electronic device, so as to facilitate use.

[0165] The above description is only exemplary embodiments of the present disclosure, and cannot limit the scope of the present disclosure. Any equivalent changes and modifications made in accordance with the teachings of the present disclosure are still within the scope of the present disclosure. Other embodiments of the present disclosure will be readily apparent to those skilled in the art upon considering the specification and practicing the present disclosure. The present application is intended to cover any variations, uses, or adaptive changes of the present disclosure that follow the general principles of the present disclosure and include common knowledge or conventional technical means in the technical field not described in the present disclosure. The scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method for detecting abnormal parking by an inspection vehicle, characterized in that, Applied to inspection vehicles equipped with cameras, the method includes: The camera acquires parking space video data, extracts parking space information of each parking space in the current frame from the parking space video data, and extracts chassis information and license plate information of each vehicle in the current frame. Based on the parking space information, update the parking space tracking information of the corresponding parking space to obtain the target parking space tracking information of the corresponding parking space; update the license plate tracking information of the corresponding vehicle based on the license plate information to obtain the target license plate tracking information of the corresponding vehicle; and update the chassis tracking information of the corresponding vehicle based on the chassis information to obtain the target chassis tracking information of the corresponding vehicle. Matching the latest license plate information in the target license plate tracking information corresponding to the latest chassis information in each of the target chassis tracking information to obtain matched chassis information, wherein the latest chassis information includes a chassis quadrilateral, the matching process includes: determining two vehicle front points corresponding to the latest chassis information; translating the two vehicle front points upward along the vertical direction of the plane containing the chassis quadrilateral to obtain a vehicle front plane; determining whether the license plate corresponding to each of the latest license plate information is located on the vehicle front plane, and if so, establishing a matching relationship between the corresponding latest license plate information and the latest chassis information to obtain matched chassis information, and determining the latest parking space information in the target parking space tracking information corresponding to the matched chassis information; Based on the relative positional relationship between the matched chassis information and the corresponding latest parking space information, determine whether the parking status of the vehicle corresponding to the latest license plate information is abnormal, including: if the relative positional relationship is within the corresponding latest parking space, determine the first parking status as normal and count the first cumulative count of the first parking status; if the relative positional relationship intersects with the corresponding latest parking space, determine the second parking status as abnormal and count the second cumulative count of the second parking status; if the relative positional relationship is not within the corresponding latest parking space, determine the third parking status as abnormal and count the third cumulative count of the third parking status; count the number of consecutive frames for which the target license plate tracking information corresponding to the matched chassis information has not been updated, and if the number of consecutive frames is greater than a preset number of frames, determine the parking status corresponding to the maximum value among the first cumulative count, the second cumulative count, and the third cumulative count as the parking status of the vehicle corresponding to the latest license plate information.

2. The method for detecting abnormal parking by an inspection vehicle according to claim 1, characterized in that, The extraction of chassis and license plate information for each vehicle in the current frame specifically includes: The current frame is input into the trained Deep3DBox algorithm to obtain the 3D bounding boxes of each vehicle in the current frame. Project the bottom of each 3D bounding box onto a 2D plane to obtain the chassis information corresponding to each vehicle; The current frame is input into the trained license plate detection model to obtain the license plate information of each vehicle.

3. The method for detecting abnormal parking by an inspection vehicle according to claim 1, characterized in that, The step of updating the berth tracking information of the corresponding berth based on the berth information to obtain the target berth tracking information of the corresponding berth specifically includes: The first berth quadrilateral corresponding to the latest berth information in the berth tracking information of each berth is used to calculate the quadrilateral IOU with the second berth quadrilateral corresponding to each berth information to obtain the IOU value; The IOU values ​​are compared with the IOU threshold to filter the berth information set where the IOU value is greater than the IOU threshold; The target berth information with the largest IOU value in the berth information set is selected and updated to the corresponding berth tracking information to obtain the target berth tracking information for the corresponding berth.

4. The method for detecting abnormal parking by an inspection vehicle according to claim 1, characterized in that, The license plate information includes the license plate number and license plate location information. Based on this license plate information, the corresponding vehicle's license plate tracking information is updated to obtain the target license plate tracking information for the corresponding vehicle. Specifically, this includes: The voting license plate number in the license plate tracking information of each vehicle is determined, and the voting license plate number is the license plate number that appears most frequently. Each of the aforementioned license plate numbers is matched character-by-character with the voting license plate number to obtain the number of matched characters; Based on the relative positional relationship between each of the license plate location information and the latest license plate location information in the license plate tracking information corresponding to the voting license plate number, the corresponding number of matching characters is adjusted to obtain the target number of matching characters; Among the license plate information with a target matching character count greater than the character count threshold, the license plate information with the largest target matching character count is selected and updated to the corresponding vehicle's license plate tracking information to obtain the target license plate tracking information for the corresponding vehicle.

5. The method for detecting abnormal parking by an inspection vehicle according to claim 4, characterized in that, The step of adjusting the number of matching characters based on the relative positional relationship between each of the license plate location information and the latest license plate location information in the license plate tracking information corresponding to the voting license plate number, to obtain the target number of matching characters, specifically includes: If the relative positions overlap, the number of corresponding matched characters is increased by a preset value to obtain the target number of matched characters; If the relative positions do not intersect, then the number of matching characters is determined as the target number of matching characters.

6. A device for detecting abnormal parking in an inspection vehicle, used to implement the method for detecting abnormal parking in an inspection vehicle as described in any one of claims 1 to 5, characterized in that, include: The information acquisition module (11) is used to acquire parking space video data through a camera, extract parking space information of each parking space in the current frame from the parking space video data, and extract chassis information and license plate information of each vehicle in the current frame. The tracking update module (12) is used to update the parking tracking information of the corresponding parking space according to the parking space information to obtain the target parking space tracking information of the corresponding parking space, update the license plate tracking information of the corresponding vehicle according to the license plate information to obtain the target license plate tracking information of the corresponding vehicle, and update the chassis tracking information of the corresponding vehicle according to the chassis information to obtain the target chassis tracking information of the corresponding vehicle. The information matching module (13) is used to match the latest license plate information in the target license plate tracking information corresponding to the latest chassis information in each target chassis tracking information to obtain the matched chassis information, and to determine the latest parking space information in the target parking space tracking information corresponding to the matched chassis information. The anomaly detection module (14) is used to determine whether the parking status of the vehicle corresponding to the latest license plate information is abnormal based on the relative positional relationship between the matched chassis information and the corresponding latest parking space information.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is loaded and executed by the processor, it employs the method described in any one of claims 1-5.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, When the processor loads and executes the computer program, it employs the method described in any one of claims 1-5.

Images