Dynamic tracking and evidence obtaining method for vehicle license plate shielding behavior of road weighing platform

Abstract

The invention relates to the field of road traffic management, in particular to a road weighing platform vehicle license plate shielding behavior dynamic tracking and evidence obtaining method. According to the method, multi-modal sensors such as a millimeter wave radar, a high-definition camera, a thermal imaging device and a laser radar are deployed, a hierarchical data processing system is constructed, vehicle driving data are collected in real time, and multi-dimensional verification analysis is carried out. When the system detects that a vehicle enters a weighing area, lightweight video preliminary screening is triggered, dual verification is carried out in combination with laser radar point cloud density analysis and thermal imaging temperature difference detection, and finally the shielding behavior is judged through a fusion judgment mechanism. The problems that a single sensor is high in misjudgment rate, low in recognition rate in a severe environment, incapable of forming a complete evidence chain and the like are solved, all-weather high-precision detection is achieved, and the law enforcement efficiency and the intelligent level are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of highway traffic management technology, specifically a method for dynamically tracking and collecting evidence of vehicle license plate obstruction on highway weighbridges. Background Technology

[0002] Currently, there are common technical bottlenecks in the field of vehicle license plate occlusion detection. Existing solutions mostly rely on a single sensor for static capture. Due to environmental adaptability, the recognition rate of license plate characters drops significantly in complex scenarios such as low light, rainy or foggy weather, or strong backlight. Furthermore, it is difficult to distinguish between normal blurring caused by dirt or reflection and human occlusion, resulting in a high false detection rate.

[0003] Traditional detection systems are typically designed only for stationary vehicles, lacking the ability to continuously track dynamically moving vehicles. When a vehicle passes through the detection area quickly, occlusion can easily be missed due to data gaps between frames. Furthermore, existing technologies mostly focus on single-point detection, failing to form a closed loop from trigger perception to evidence fixation. After detecting occlusion, they cannot automatically link it to the vehicle's historical trajectory, real-time driving path, and surrounding monitoring data, making it difficult to construct a complete chain of evidence of violations. This often leads to disputes during law enforcement due to insufficient correlation of evidence.

[0004] In the evidence collection process, traditional equipment often uses distributed storage, making the original data easy to be tampered with or lost. Furthermore, it lacks an efficient cross-regional collaboration mechanism, resulting in low efficiency in tracking illegal vehicles and making it difficult to meet the demands of modern traffic management for intelligence and real-time processing. Summary of the Invention

[0005] The purpose of this invention is to provide a method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior on highway weighbridges, achieving all-weather monitoring through multimodal sensor fusion. The system deploys millimeter-wave radar, high-definition cameras, thermal imaging equipment, and lidar, constructing a hierarchical data processing mechanism: in the detection triggering stage, millimeter-wave radar tracks vehicle trajectories in real time; in the lightweight detection stage, video is used for rapid initial screening of suspected obstructed vehicles; in the multimodal verification stage, lidar point cloud density analysis and thermal imaging temperature difference detection are combined for dual verification; finally, the obstruction behavior is comprehensively determined through a fusion judgment mechanism. This effectively solves the problems of traditional single-sensor detection being susceptible to environmental interference, having a high false detection rate, and being unable to adapt to complex scenarios, thus addressing the issues raised in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for dynamically tracking and collecting evidence of vehicle license plate obstruction includes the following steps:

[0008] Step 1: Deploy various types of sensor devices to collect vehicle driving data, including millimeter-wave radar, high-definition panoramic camera, close-up supplementary lighting camera, thermal imaging camera and lidar;

[0009] Step 2: Perform hierarchical data processing on video data, point cloud data and thermal imaging data collected by various sensor devices, and complete the detection triggering stage, lightweight detection stage, multimodal verification stage and fusion judgment stage in sequence;

[0010] Step 3: Based on the results of the fusion judgment stage, trigger the automatic evidence collection operation to collect spatiotemporal data of vehicle occlusion behavior;

[0011] Step 4: Encrypt and transmit the collected spatiotemporal data to the cloud platform. The cloud platform will conduct cross-regional tracking and evidence chain integration. The evidence chain must include at least the vehicle's trajectory across the weighbridge, the obstruction report from multi-sensor analysis, and the vehicle registration information, and form an evidence package.

[0012] Step 5: Send the evidence package to the information application platform, which provides services for viewing and retrieving the evidence package.

[0013] Preferably, the deployment method for the sensor device in step one is as follows:

[0014] A combination of high-definition panoramic cameras and close-up supplementary lighting cameras was installed above the road 50 meters in front of and behind the weighing platform.

[0015] Millimeter-wave radar is installed 10 meters in front of and behind the weighing platform, with a detection radius of ≥200 meters;

[0016] Thermal imaging cameras were installed at the entrance area of ​​the weighing platform;

[0017] A lidar is fixed directly above the weighing platform, and its scanning range covers the three-dimensional outline of the entire vehicle.

[0018] Preferably, the specific execution for the detection triggering phase in step two is as follows:

[0019] The millimeter-wave radar continuously scans the road ahead, and when it detects a vehicle entering its monitoring range, it begins to track and record the vehicle's position, speed, and trajectory.

[0020] When the millimeter-wave radar determines that a vehicle has entered the 20-meter warning range of the weighing area of ​​the scale, it triggers the high-definition camera to switch to 4K high-definition mode.

[0021] Simultaneously activate the lidar and thermal imaging camera to collect multi-dimensional data;

[0022] The collected video data, point cloud data, and thermal imaging data are uploaded to the hierarchical computing module and then enter the lightweight detection stage.

[0023] Preferably, the lightweight detection stage of step two is specifically implemented as follows:

[0024] Analyze video data to quickly locate license plate areas;

[0025] Further analysis and testing of the license plate area are conducted to determine the integrity of the license plate and the matching between the license plate and the vehicle.

[0026] If no license plate is detected, the license plate does not match the standard format, the recognized license plate characters are blurry or missing, or the license plate recognition result contradicts the actual characteristics of the vehicle, it is initially judged as suspected occlusion and enters the multimodal verification stage. This stage only uses video data to quickly screen highly suspicious targets, avoiding comprehensive detection of all vehicles, which can significantly reduce storage and computing costs and greatly improve system operating efficiency.

[0027] Preferably, the following specific steps are performed for the multimodal verification stage in step two:

[0028] The point cloud density of the LiDAR is analyzed, and the threshold for the difference in point cloud density in the vehicle body area is set to 45%. If the point cloud density is significantly different from that of other areas of the vehicle body and exceeds 45%, it indicates that there is a significant covering in that area, and it is marked as a suspected LiDAR obstruction.

[0029] Thermal imaging temperature difference detection is performed, and a temperature anomaly judgment range of ±3℃ is established. If the temperature of this area deviates significantly from the temperature around the vehicle and exceeds 3℃, and the shape conforms to the characteristics of common obstructions, it is marked as a suspected thermal imaging obstruction.

[0030] Preferably, the specific execution for the fusion judgment stage in step two is as follows:

[0031] Integrating suspected occlusion markers from video data recognition, LiDAR suspected occlusion markers, and thermal imaging suspected occlusion markers;

[0032] Occlusion detection is triggered when the number of abnormal markers is greater than or equal to 2.

[0033] Preferably, the specific execution of the automated evidence collection operation in step three is as follows:

[0034] A high-definition panoramic camera captures clear images containing the time, location, and weighing platform number watermark, and records a 10-second video of the key process, including a close-up of the license plate area 3 seconds before the vehicle enters the weighing platform, the data of the whole vehicle and the weighing platform during the 4-second weighing process, and a close-up of the license plate area again 3 seconds after the vehicle leaves.

[0035] A close-up supplemental lighting camera captures high-definition images of the license plate area.

[0036] Preferably, the specific implementation of step four, cross-regional tracking and evidence chain integration, is as follows:

[0037] By matching vehicle trajectory, the detection data of the current weighing platform is associated with the historical records of the vehicle to reconstruct the cross-weighing platform trajectory map and global driving path of the obstructing vehicle.

[0038] Connect to the public security traffic management database to retrieve vehicle registration information, including vehicle owner, vehicle model, and contact information;

[0039] By combining the obstruction reports from multi-sensor analysis, a complete chain of evidence is generated, and the data is packaged to form a complete evidence package.

[0040] Preferably, the specific implementation for the information application platform in step five is as follows:

[0041] Receive and visually display evidence packets sent by the automatic evidence collection module;

[0042] The evidence package is provided for law enforcement officers to retrieve, and the evidence package is manually reviewed. Once the obstruction is confirmed, a penalty notice is issued.

[0043] This allows car owners to view their traffic violation records and detailed violation information.

[0044] Preferably, the local end of the automatic evidence collection module only retains key metadata, including the judgment result, timestamp, and sensor raw data index, as well as low-resolution thumbnails, to reduce storage pressure.

[0045] Compared with the prior art, the beneficial effects of the present invention are:

[0046] 1. This invention constructs a dynamic monitoring system that integrates multimodal perception and step-by-step verification. By fusing the precise trajectory tracking of millimeter-wave radar, the visual capture of high-definition cameras, the temperature feature analysis of thermal imaging, and the three-dimensional point cloud modeling of lidar, it achieves multi-dimensional three-dimensional monitoring of vehicle license plate occlusion behavior. In the multimodal verification stage, point cloud density and temperature difference are combined for dual verification. Finally, the occlusion behavior is comprehensively judged through a fusion judgment mechanism, which effectively avoids the risk of misjudgment by a single sensor, greatly improves the recognition accuracy, and can adapt well to harsh weather or complex lighting conditions, always maintaining a high recognition rate.

[0047] 2. By adopting a hierarchical processing mechanism, this invention activates only the necessary sensors during the triggering phase, effectively reducing overall power consumption. During the lightweight detection phase, video screening is used to reduce computational load, thereby effectively improving the daily average vehicle license plate processing capacity of the lane.

[0048] 3. This invention, through cross-platform data linkage in the evidence collection process, not only ensures the immutability of the evidence chain, but also enables precise tracing of illegal acts by linking to the public security database, thereby improving the completeness of the evidence chain and effectively reducing the administrative review rate. Attached Figure Description

[0049] Figure 1This is a schematic diagram of the system structure of the present invention;

[0050] Figure 2 This is a schematic diagram of the system flow of the present invention. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] To address the issues of high false positive rates, low recognition rates in harsh environments, and inability to form a complete tracking and evidence collection chain in existing vehicle license plate obscuring technologies that rely on a single sensor, please refer to [link to relevant documentation]. Figure 1-2 This embodiment provides the following technical solution:

[0053] A method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior on highway weighbridges. This method is based on a process-oriented system that includes multiple types of sensor devices, hierarchical calculation modules, automatic evidence collection modules, and information application platforms.

[0054] Multiple types of sensor devices serve as the front end of the system, deployed in the highway weighbridge area for comprehensive environmental and vehicle monitoring. The specific configuration of these sensor devices is as follows:

[0055] High-definition panoramic cameras and close-up supplementary lighting cameras are installed 50 meters above the road in front of and behind the weighing platform. The high-definition panoramic cameras are used to collect images and video data of the front license plate, rear license plate, and side areas of the vehicle. The high-definition panoramic cameras use 8mm focal length lenses and can achieve a minimum resolution of 50 pixels in the license plate area at a distance of 50 meters, which meets the ANPR system recognition standard. The close-up supplementary lighting cameras have built-in infrared supplementary lights and white light strobe devices to ensure that clear images and videos can still be obtained in adverse environments such as low light, rain, and fog.

[0056] Millimeter-wave radars are deployed 10 meters in front of and behind the weighing platform to output the vehicle's position, speed, and trajectory in real time. The detection radius of the millimeter-wave radars is at least 200 meters. This value is based on experimental verification: a detection distance of 200 meters can ensure at least 6 seconds of warning time for vehicles with speeds of 0-120 km / h, meeting the system response requirements.

[0057] A thermal imaging camera is installed at the entrance area of ​​the weighing platform. The thermal imaging camera is used to collect thermal imaging data of the vehicle and analyze the material properties of the obstruction.

[0058] A lidar is mounted on a bracket directly above the weighing platform to scan the vehicle's 3D contours and collect its 3D point cloud data.

[0059] All monitoring equipment, including high-definition panoramic cameras, close-up supplementary lighting cameras, millimeter-wave radar, thermal imaging cameras, and lidar, are connected to the hierarchical computing module via a wired network to ensure low latency in data transmission. A precision clock protocol is used to achieve microsecond-level time synchronization, and a calibration board is used to uniformly calibrate the spatial coordinates of multiple sensors to ensure spatiotemporal consistency of data.

[0060] The hierarchical calculation module serves as the processing middleware of the system. Its execution process can be divided into a detection triggering phase, a lightweight detection phase, a multimodal verification phase, and a fusion judgment phase, as specifically configured below:

[0061] During the detection triggering phase, the millimeter-wave radar continuously scans the road ahead. When a vehicle enters its monitoring range and is within 200 meters of the weighing platform, the millimeter-wave radar begins to track and record its position, speed, and trajectory. When it determines that the vehicle is about 20 meters away from the weighing area of ​​the platform, the millimeter-wave radar sends a trigger signal to the step-by-step calculation module, indicating that the vehicle is about to enter the weighing area. After receiving the signal, the step-by-step calculation module immediately activates the high-definition panoramic camera and switches to 4K high-definition mode. At the same time, it starts the lidar and thermal imaging camera to scan the vehicle synchronously. The collected video data, point cloud data, and thermal imaging data are uploaded to the step-by-step calculation module and then enter the lightweight detection phase.

[0062] In the lightweight detection phase, the step-by-step computation module first analyzes only the video data. A YOLOv5-based object detection model quickly locates the license plate region. This model preprocesses video frames, extracts features, and performs bounding box regression, outputting the license plate coordinates. Subsequently, OCR technology is used to recognize the license plate characters and match them against a standard format. If no license plate is detected, the license plate does not match the standard format, the recognized license plate characters are blurry, characters are missing, or the license plate recognition result contradicts the actual characteristics of the vehicle, it is initially judged as a suspected occlusion, and the system proceeds to the multimodal verification phase. This phase quickly filters highly suspicious targets based solely on video data, avoiding comprehensive detection of all vehicles, significantly reducing storage and computational costs, and greatly improving system operating efficiency.

[0063] In the multimodal verification phase, the hierarchical calculation module further analyzes the point cloud data of the LiDAR and the data of the thermal imaging camera. After preprocessing, the point cloud data is segmented into the license plate area using a region growing algorithm. The difference between its point cloud density and other areas of the vehicle body is calculated, and the three-dimensional contour of the license plate area is extracted. The threshold for the difference in point cloud density in the vehicle body area is set to 45% (based on experimental statistics: the density fluctuation of the normal license plate area is less than 30%, and the density decrease due to occlusion exceeds 45%). If the point cloud density is significantly different from other areas of the vehicle body and exceeds 45%, it indicates that there is a significant covering in the area, and it is marked as a suspected occlusion of the LiDAR.

[0064] Meanwhile, the thermal imaging data uses an adaptive threshold segmentation method to automatically identify the license plate area, extract temperature distribution, and capture the temperature difference between the license plate area and the vehicle body. The deviation threshold is set at 3℃ (based on actual measurement data: the temperature difference between the license plate and the vehicle body is usually ≤2℃ when the vehicle is running, and the temperature difference is ≥3℃ due to obstructions such as cloth or plastic). If the temperature of this area deviates significantly from the temperature around the vehicle and exceeds 3℃, there may be suspected obstruction. At the same time, the shape of the abnormal temperature area is judged, and the features of common obstructions are matched by contour extraction (such as rectangular contour, aspect ratio 1:2~1:3, area 100-500 pixels). If the shape matches the features of common obstructions, it is marked as suspected thermal imaging obstruction. The temperature difference helps to distinguish the type of obstruction, providing detailed support for subsequent evidence collection.

[0065] During the fusion and judgment phase, the step-by-step calculation module performs a comprehensive judgment based on suspected occlusion markers identified from video data, LiDAR, and thermal imaging. Specifically, if any two or all three criteria are met—video recognition of an abnormal license plate area, LiDAR detection of abnormal point cloud density, or thermal imaging detection of abnormal temperature distribution—the vehicle is determined to be a "license plate occlusion vehicle." For example, if the video shows the license plate completely covered and the point cloud density in the license plate area is reduced by 70%, it is determined to be occluded; or if the video shows blurred license plate characters and thermal imaging shows the temperature in that area is 3°C lower than the vehicle body, it is also determined to be occluded. This fusion mechanism avoids misjudgment from a single sensor and improves the accuracy of the judgment.

[0066] The automatic evidence collection module, as the system's backend, is used to automatically perform evidence collection operations after the hierarchical calculation module determines that the vehicle is a "license plate obscured vehicle." The specific settings are as follows:

[0067] A high-definition panoramic camera captures clear images containing the time, location, and weighing platform number watermark, and records 10 seconds of high-definition video. This includes a close-up of the license plate area 3 seconds before the vehicle enters the weighing platform, the data of the whole vehicle and the weighing platform during the 4-second weighing process, and a close-up of the license plate area again 3 seconds after the vehicle leaves. Combined with the 3D outline of the vehicle captured by LiDAR, multiple feature points of the vehicle whose license plate is obscured are extracted, including brand and model, body color, body outline, window shape, headlight position, and annual inspection mark. These feature points are combined to form a temporary unique identifier for the vehicle.

[0068] At the same time, the close-up supplementary lighting camera takes a local high-definition picture of the license plate area to supplement and record the obscured details;

[0069] All initial data is encrypted and uploaded to the cloud platform. The local device only retains key metadata, including the judgment result, timestamp, and sensor raw data index, as well as low-resolution thumbnails, to reduce storage pressure.

[0070] After receiving the initial data, the cloud platform conducts cross-regional tracking and evidence chain integration. Based on timestamps, the cloud platform aligns data from multiple weighbridges, integrates GPS coordinates and millimeter-wave radar trajectories, uses Kalman filtering to predict vehicle paths, and extracts historical records stored in a distributed database to reconstruct the global driving path of the obscured vehicle. Simultaneously, it connects to the public security traffic management database, compares the feature points of the vehicle with the registered vehicles in the database, thereby identifying the vehicle with the obscured license plate, retrieving vehicle registration information including the owner, vehicle model, and contact information, forming a complete evidence chain including the vehicle's cross-weighbridge trajectory map, obstruction reports from multi-sensor analysis, and vehicle registration information, and then packages the data into a complete evidence package.

[0071] The evidence package is stored using blockchain technology to ensure the immutability of the data, thereby guaranteeing the fairness and compliance of law enforcement. After the evidence package is formed, the cloud platform sends it to the information application platform for personnel to retrieve and view.

[0072] The information application platform also serves as the system backend, receiving and displaying evidence packages sent by the automatic evidence collection module. On the one hand, law enforcement officers retrieve the evidence packages through the application platform during law enforcement, conduct a final manual review, and issue a penalty notice after confirming that the obstruction behavior is correct. The notice is then sent to the vehicle owner via SMS or navigation apps, enabling the owner to handle the matter promptly. On the other hand, vehicle owners can also log in to the information application platform to view their violation records and detailed violation information.

[0073] Working Principle: The system achieves all-weather environmental perception through multiple sensor devices deployed in the weighing platform area. When a vehicle enters the 200-meter monitoring range of the millimeter-wave radar, the radar begins to track the vehicle's trajectory and calculate its speed. When the vehicle is approximately 20 meters away from the weighing area, the radar triggers the high-definition panoramic camera to switch to high-definition mode, simultaneously activating the lidar and thermal imaging camera to synchronously collect vehicle data. The collected video stream, 3D point cloud data, and thermal imaging information are transmitted via a wired network to a series of computing modules for processing.

[0074] The hierarchical calculation module employs a tiered processing strategy. First, lightweight video analysis quickly locates the license plate area. If an anomaly is detected, it is marked as suspected occlusion and enters the multimodal verification stage. In this stage, point cloud data generated by LiDAR is used to construct the vehicle's 3D contour. By comparing the point cloud density of the license plate area with other areas of the vehicle body, the presence of any covering is identified. Simultaneously, a thermal imaging camera captures the temperature distribution characteristics of the license plate area, analyzing its temperature difference and temperature field morphology with the surrounding vehicle body to help determine the material of the occlusion. Only when any two of the three indicators—video analysis, LiDAR point cloud detection, and thermal imaging temperature analysis—are abnormal will the system ultimately determine that the vehicle has an occluded license plate.

[0075] Once the obstruction is confirmed, multi-dimensional evidence collection is initiated immediately. All raw data is encrypted and uploaded to the cloud platform. The cloud platform uses blockchain technology to store the evidence package and links it to the vehicle's historical trajectory and public security database information to build a complete chain of evidence, including the vehicle's crossing the weighbridge, the obstruction analysis report, and the vehicle owner's information.

[0076] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0077] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

Claims

1. A method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior, characterized in that, Includes the following steps: Step 1: Deploy various types of sensor devices to collect vehicle driving data, including millimeter-wave radar, high-definition panoramic camera, close-up supplementary lighting camera, thermal imaging camera and lidar; Step 2: Perform hierarchical data processing on video data, point cloud data and thermal imaging data collected by various sensor devices, and complete the detection triggering stage, lightweight detection stage, multimodal verification stage and fusion judgment stage in sequence; Step 3: Based on the results of the fusion judgment stage, trigger the automatic evidence collection operation to collect spatiotemporal data of vehicle occlusion behavior; Step 4: Encrypt and transmit the collected spatiotemporal data to the cloud platform. The cloud platform will conduct cross-regional tracking and evidence chain integration. The evidence chain must include at least the vehicle's trajectory across the weighbridge, the obstruction report from multi-sensor analysis, and the vehicle registration information, and form an evidence package. Step 5: Send the evidence package to the information application platform, which provides services for viewing and retrieving the evidence package.

2. The method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior according to claim 1, characterized in that, The specific deployment method for the sensor equipment in step one is as follows: A combination of high-definition panoramic cameras and close-up supplementary lighting cameras was installed above the road 50 meters in front of and behind the weighing platform. Millimeter-wave radar is installed 10 meters in front of and behind the weighing platform, with a detection radius of ≥200 meters; Thermal imaging cameras were installed at the entrance area of ​​the weighing platform; A lidar is fixed directly above the weighing platform, and its scanning range covers the three-dimensional outline of the entire vehicle.

3. The method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior according to claim 1, characterized in that, Specifically, the detection triggering phase in step two is executed as follows: The millimeter-wave radar continuously scans the road ahead, and when it detects a vehicle entering its monitoring range, it begins to track and record the vehicle's position, speed, and trajectory. When the millimeter-wave radar determines that a vehicle has entered the 20-meter warning range of the weighing area of ​​the scale, it triggers the high-definition camera to switch to 4K high-definition mode. Simultaneously activate the lidar and thermal imaging camera to collect multi-dimensional data; The collected video data, point cloud data, and thermal imaging data are uploaded to the hierarchical computing module and then enter the lightweight detection stage.

4. The method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior according to claim 1, characterized in that, Specifically, the lightweight detection phase in step two will be performed as follows: The video data is analyzed, and the license plate area is quickly located using an object detection model based on the YOLOv5 algorithm. The video frames are preprocessed, features are extracted, and bounding box regression is performed to output the license plate coordinates. Use OCR technology to recognize license plate characters and match them with a standard format; If no license plate is detected at all, the license plate cannot match the standard format, the recognized license plate characters are blurry or missing, or the license plate recognition result contradicts the actual characteristics of the vehicle, it is initially determined to be suspected occlusion and enters the multimodal verification stage.

5. The method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior according to claim 1, characterized in that, The specific execution of step two, the multimodal verification phase, is as follows: The point cloud density of the LiDAR is analyzed, and the threshold for the difference in point cloud density in the vehicle body area is set to 45%. If the threshold is exceeded, it is marked as a suspected occlusion of the LiDAR. Thermal imaging temperature difference detection is performed, and a temperature anomaly judgment range of ±3℃ is established. If the temperature exceeds the judgment range and the shape matches the characteristics of common obstructions, it is marked as a suspected thermal imaging obstruction.

6. The method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior according to claim 1, characterized in that, Specifically, the fusion judgment phase in step two is executed as follows: Integrating suspected occlusion markers from video data recognition, LiDAR suspected occlusion markers, and thermal imaging suspected occlusion markers; Occlusion detection is triggered when the number of abnormal markers is greater than or equal to 2.

7. The method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior according to claim 1, characterized in that, The specific execution of the automated evidence collection operation in step three is as follows: A high-definition panoramic camera captures clear images containing the time, location, and weighing platform number watermark, and records a 10-second video of the key process. A close-up supplemental lighting camera captures high-definition images of the license plate area.

8. The method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior according to claim 1, characterized in that, The specific implementation of step four, cross-regional tracking and evidence chain integration, is as follows: By matching vehicle trajectory, the detection data of the current weighing platform is associated with the historical records of the vehicle to reconstruct the cross-weighing platform trajectory map and global driving path of the obstructing vehicle. Connect to the public security traffic management database to retrieve vehicle registration information, including vehicle owner, vehicle model, and contact information; By combining the obstruction reports from multi-sensor analysis, a complete chain of evidence is generated, and the data is packaged to form a complete evidence package.

9. A method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior according to claim 1, characterized in that, The specific implementation of step five, the information application platform, is as follows: Receive and visually display evidence packets sent by the automatic evidence collection module; The evidence package is provided for law enforcement officers to retrieve, and the evidence package is manually reviewed. Once the obstruction is confirmed, a penalty notice is issued. This allows car owners to view their traffic violation records and detailed violation information.

10. A method for dynamic tracking and evidence collection of vehicle license plate obstruction behavior according to claim 1, characterized in that, The local end of the automatic evidence collection module only retains key metadata, including the judgment result, timestamp, and sensor raw data index, as well as low-resolution thumbnails.

Images