Truck red light running snapshot method and device based on image recognition

Abstract

The invention relates to the technical field of video monitoring, and discloses a lorry red light running snapshot method and device based on image recognition, and the method achieves the omnibearing monitoring of a traffic intersection through capturing the image data of a lorry passing through the traffic intersection and dynamically adjusting the parameters of the snapshot device and the image capturing frequency through an image capturing strategy. The problems of blind areas and missing detection are avoided; image preprocessing is carried out on captured image data, so that the accuracy of image recognition is improved, and the influence of noise and interference on target detection is reduced; target detection is carried out on the truck in the image data, the existence of the truck is judged, the problems of multi-target detection and target tracking are solved, and it is ensured that the truck is accurately recognized; performing behavior analysis on the detected truck, identifying a red light running behavior of the truck, if the truck has the red light running behavior, marking the image data as a violation image, storing the violation image in a database, and performing privacy protection on the violation image; and if a violation image is monitored, alarm processing is automatically triggered.

Description

Technical Field

[0001] The present invention relates to the technical field of video monitoring, and in particular to a method and device for capturing a truck running a red light based on image recognition. Background Art

[0002] The current research is to use deep learning algorithms for target detection and behavior analysis, and to improve the accuracy and reliability of truck violation capture through multi-angle camera layout and infrared camera technology. Truck violation capture can automatically detect and record violations, reducing the need for manual intervention. Through real-time data processing and network collaboration, it can achieve timely monitoring and alarm processing of truck violations. However, most of them have not solved how to configure the capture device to capture image data and perform target detection on trucks, conduct behavior analysis after determining the presence of trucks, and mark the violation image and automatically trigger alarm processing after identifying the truck running a red light.

[0003] For example, the Chinese patent with publication number CN115762176A discloses a method, device, computer equipment, computer-readable storage medium and computer program product for detecting and capturing red light running behavior. The method determines the time change law of the current traffic light at the intersection according to the historical traffic light change time at the intersection; predicts the time period when the subsequent traffic light at the intersection is red based on the traffic light time change law; and controls the camera to start red light running behavior detection and capture during the time period when the prediction result indicates that the traffic light is red. By learning the time change law of the traffic light at the current intersection, and predicting the time period when the subsequent traffic light at the intersection is red based on the traffic light time change law, and controlling the camera to start red light running behavior detection and capture during the time period when the traffic light is red, the problem of false capture or missed capture of the camera in the detection and capture of red light running behavior is effectively solved.

[0004] For example, a Chinese patent with authorization announcement number CN102110369B discloses a method and device for capturing red light running. A high-definition camera and an embedded digital signal processor are used to sequentially capture high-definition images, and the contents of the high-definition images are used to identify red light running and calculate the vehicle speed. When a valid red light running event is confirmed, three high-definition images are combined into one red light running image, which is then transmitted to a remote server via a network. The method and system described in the invention can effectively avoid mis-shooting and missed shots, and can especially clearly identify license plates and traffic light status, further and more significantly reduce network burden and save remote server and database resources.

[0005] The above patents have the problems raised by this background technology: the above red light running behavior detection and capture method and device determine the time change law of the current traffic light at the traffic intersection according to the traffic light change time of the traffic light at the traffic intersection, predict the time period when the subsequent traffic lights at the traffic intersection are red based on the traffic light time change law, and control the camera to start red light running behavior detection and capture; the above red light running capture method and capture device use a high-definition camera and an embedded digital signal processor to sequentially shoot high-definition pictures and perform red light running recognition on the content of the high-definition pictures and calculate the vehicle speed. When it is confirmed to be a valid red light running time, the three high-definition pictures are combined into one red light running picture. The above two patents do not solve the problem of how to capture image data and perform target detection on trucks by configuring a capture device, perform behavior analysis after determining the existence of a truck, and mark the illegal image after identifying that the truck has run a red light and automatically trigger an alarm process. To solve this problem, the present invention proposes a truck running a red light capture method and device based on image recognition. Summary of the invention

[0006] The purpose of this section is to summarize some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0007] In view of the problems existing in the above-mentioned existing methods and devices for capturing trucks running red lights based on image recognition, the present invention is proposed.

[0008] Therefore, the object of the present invention is to provide a method and device for capturing a truck running a red light based on image recognition.

[0009] In order to solve the above technical problems, the present invention provides a method for capturing a truck running a red light based on image recognition: capturing image data of a truck passing through a traffic intersection, and dynamically adjusting the parameters of a capturing device and the image capturing frequency through an image capturing strategy;

[0010] Performing image preprocessing on the captured image data;

[0011] Perform target detection on trucks in the image data to determine the existence of trucks;

[0012] Conduct behavioral analysis on detected trucks to identify red light running behaviors. If a truck runs a red light, the image data will be marked as a violation image and saved in the database. At the same time, the privacy of the violation image will be protected.

[0013] If an illegal image is detected, an alarm will be automatically triggered.

[0014] As a preferred solution of the method for capturing a truck running a red light based on image recognition according to the present invention, the image data of the truck passing through the traffic intersection is captured, and the parameters of the capturing device and the image capturing frequency are dynamically adjusted through the image capturing strategy, and the image capturing strategy includes:

[0015] According to the size and shape of the traffic intersection, the number and location of the capture devices are determined, and the height and angle are configured for each capture device. The resolution of the camera of each capture device is configured according to the requirements of truck identification, the frame rate is selected according to the storage requirement relationship, and the exposure time is automatically adjusted according to the lighting conditions. According to the real-time traffic flow and the preset time interval, the working state of the capture device is dynamically adjusted, and the traffic flow threshold is configured. If the real-time traffic flow is greater than the traffic flow threshold, the capture frequency of the capture device is increased; if the real-time traffic flow is equal to the traffic flow threshold, the capture device captures the image data at the preset time interval; if the real-time traffic flow is less than the traffic flow threshold, the capture frequency of the capture device is reduced.

[0016] As a preferred solution of the method for capturing a truck running a red light based on image recognition according to the present invention, the truck in the image data is detected by a target detection strategy, and the target detection strategy includes:

[0017] The image data is divided into regional blocks, and the scanning window is configured according to the size of the truck. The regional blocks are scanned and identified through the scanning window, and the regional blocks that do not contain the truck are excluded. At the same time, the truck features are extracted in each regional block. The existence of the truck in the regional block is judged by the classification decision function, and the overlapping regional windows are obtained. The overlapping degree of each regional window is calculated to obtain the regional window where the truck is located, and the regional window where the truck is detected is labeled.

[0018] As a preferred solution of the method for capturing trucks running red lights based on image recognition according to the present invention, the calculation formula for the degree of overlap of each area window is as follows:

[0019] ;

[0020] In the formula, Display area window and area window The degree of overlap, Display area window and area window The area of ​​the intersection of Display area window and area window The area of ​​the union of .

[0021] As a preferred solution of the method for capturing trucks running red lights based on image recognition according to the present invention, the behavior of the detected trucks is analyzed by a behavior analysis strategy to identify the trucks running red lights. The behavior analysis strategy includes:

[0022] By obtaining the regional window of the truck in different image data, tracking the position of the truck, and calculating the speed of the truck, the state of the traffic light is determined. Combining the speed of the truck and the state of the traffic light, a comprehensive judgment is made on the truck's red light running behavior.

[0023] As a preferred solution of the method for capturing a truck running a red light based on image recognition according to the present invention, the calculation formula of the speed of the truck is as follows:

[0024] ;

[0025] In the formula, represents the speed of the truck, Indicates time The location of the truck in the corresponding image data, Indicates time The location of the truck in the corresponding image data, Indicates the time difference between different image data.

[0026] As a preferred solution of the method for capturing a truck running a red light based on image recognition according to the present invention, the comprehensive judgment of the truck running a red light includes:

[0027] If the truck crosses the stop line when the traffic light is red and the speed of the truck remains unchanged, the truck is considered to have run a red light.

[0028] If the truck's position and speed indicate that it cannot stop before the traffic light turns red when the traffic light is green or yellow, the truck is considered to have run a red light.

[0029] If the truck's position and speed show that it stops before the traffic light turns red when the traffic light is green or yellow, then the truck does not run a red light.

[0030] The device for capturing trucks running red lights based on image recognition comprises: a camera for capturing image data of trucks passing through traffic intersections.

[0031] A computer device includes a memory for storing instructions and a processor for executing the instructions, so that the computer device implements a method for capturing a truck running a red light based on image recognition.

[0032] A computer-readable storage medium stores a computer program, which, when executed, implements a method for capturing trucks running red lights based on image recognition.

[0033] The beneficial effects of the present invention are as follows: the present invention realizes all-round monitoring of traffic intersections by capturing image data of trucks passing through traffic intersections, and dynamically adjusts the parameters of the capture device and the image capture frequency through the image capture strategy, thereby avoiding the problems of blind spots and missed detections; performs image preprocessing on the captured image data to improve the accuracy of image recognition and reduce the influence of noise and interference on target detection; performs target detection on trucks in the image data to determine the existence of trucks, solves the problems of multi-target detection and target tracking, and ensures that trucks are accurately identified; performs behavioral analysis on the detected trucks to identify the red light running behavior of the trucks. If the truck runs a red light, the image data is marked as an illegal image and saved in a database. At the same time, the privacy of the illegal images is protected, and it is determined whether the truck is illegal through behavioral characteristics and rules, thereby accurately marking the illegal images; if an illegal image is detected, an alarm process is automatically triggered to achieve timely alarm and punishment, thereby improving the effect of traffic safety management.

[0034] According to the above steps, automatic monitoring and identification of trucks running red lights can be achieved, the accuracy and real-time performance of traffic violation detection can be improved, thereby effectively improving traffic safety issues. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0036] Figure 1 A method flow chart of a method for capturing a truck running a red light based on image recognition according to the present invention;

[0037] Figure 2 This is a flow chart of an image capture strategy for a truck running a red light based on image recognition according to the present invention;

[0038] Figure 3 This is a flow chart of the target detection strategy of the method for capturing a truck running a red light based on image recognition of the present invention;

[0039] Figure 4 This is a flow chart of the behavior analysis strategy of the method for capturing trucks running red lights based on image recognition according to the present invention. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0041] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0042] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0043] Example 1

[0044] In this embodiment, a method flow chart of a method for capturing a truck running a red light based on image recognition is provided, such as Figure 1 As shown, the method for capturing a truck running a red light based on image recognition includes:

[0045] S1. Capture image data of trucks passing through traffic intersections, and dynamically adjust the parameters of the capture device and the image capture frequency through image capture strategies to ensure that traffic intersections can be monitored in all directions.

[0046] Capture image data of trucks passing through traffic intersections, and dynamically adjust the parameters of the capture device and the image capture frequency through image capture strategies. Image capture strategies are as follows: Figure 2 As shown, specifically including:

[0047] Determine the number and location of capture devices according to the size and shape of the traffic intersection to ensure that every corner and driving direction is covered by at least one capture device. Configure the height and angle of each capture device to obtain a suitable viewing angle and coverage to avoid blind spots. Configure the resolution of the camera of each capture device according to the needs of truck identification to ensure that the details of the truck, including the license plate number and model, can be clearly captured. Select the frame rate based on the storage demand relationship to balance the quality of image data and storage requirements. A high frame rate can provide smoother video, but it will also take up more storage space. Automatically adjust the exposure time according to lighting conditions, including night and backlight conditions. Dynamically adjust the working state of the capture device according to the real-time traffic flow and the preset time interval. Configure the traffic flow threshold. If the real-time traffic flow is greater than the traffic flow threshold, increase the capture frequency of the capture device; if the real-time traffic flow is equal to the traffic flow threshold, the capture device captures the image data at the preset time interval; if the real-time traffic flow is less than the traffic flow threshold, reduce the capture frequency of the capture device.

[0048] The function expression for storing the demand relationship is as follows:

[0049] ;

[0050] In the formula, represents the storage requirements of image data, Indicates the frame rate of the image data, Indicates the resolution of the camera. Indicates the storage duration of the image data, Indicates the compression ratio of image data.

[0051] The frame rate of image data directly affects the storage requirements. A high frame rate means that more frames need to be stored per second. The resolution determines the amount of data in each frame. A high resolution means that each frame contains more pixels, which increases the storage requirements. The storage duration is the duration of capturing image data at preset time intervals. The longer the time, the more total frames need to be stored. The compression ratio is an adjustment factor. The compression ratio reduces the storage requirements due to high frame rate and high resolution, but the choice of compression ratio requires a balance between image quality and storage space.

[0052] The function expression for automatically adjusting the exposure time according to the lighting conditions is as follows:

[0053] ;

[0054] In the formula, Indicates the exposure time, represents the target brightness, Indicates lighting conditions.

[0055] Multiple capture devices are installed at traffic intersections. The capture device uses a 2MP telephoto camera to capture license plates within 50 meters, and a 4MP wide-angle camera to detect vehicle behavior. At the same time, the capture device is equipped with high-precision GPS and BD dual-mode, which can achieve an error of less than 5cm, ensuring that all cameras are synchronized in time to facilitate subsequent image data fusion and analysis.

[0056] According to the ambient lighting conditions and the specific requirements of the monitoring scene, the parameters of the camera in the capture device, such as exposure time, gain and focal length, are dynamically adjusted to ensure that high-quality image data can be obtained under different lighting and weather conditions, and the capture frequency of the capture device is dynamically adjusted according to the real-time traffic flow and time period.

[0057] S2. Perform image preprocessing on the captured image data to improve the accuracy of image recognition.

[0058] The captured image data is preprocessed by using Gaussian filtering to reduce random noise in the image data. The features of the truck in the image data are made more obvious by adjusting the brightness and contrast in the image data. The brightness distribution of the image data is adjusted to make the histogram of the image data more uniform, thereby enhancing the features of the truck in the image data.

[0059] Among them, the function expression for making the histogram of image data more uniform is as follows:

[0060] ;

[0061] In the formula, Indicates image data The pixel value after grayscale mapping, Indicates The number of gray levels, Represents the total number of pixels in the image data.

[0062] The contrast of the image data is enhanced according to the above formula, especially in the case of uneven lighting or bad weather conditions, which can greatly improve the visual effect of the image data, making the distinction between the truck and the background more obvious, such as the road surface and traffic lights, thereby providing clearer image input for subsequent target detection, such as truck detection and vehicle recognition.

[0063] S3. Perform target detection on the truck in the image data to determine the existence of the truck.

[0064] The target detection strategy is used to detect the truck in the image data. The target detection strategy is as follows: Figure 3 As shown, specifically including:

[0065] The image data is divided into regional blocks, and the scanning window is configured according to the size of the truck. The regional blocks are scanned and identified through the scanning window, and the regional blocks that do not contain the truck are excluded. At the same time, the truck features are extracted in each regional block. The existence of the truck in the regional block is judged by the classification decision function, and the overlapping regional windows are obtained. The overlapping degree of each regional window is calculated to obtain the regional window where the truck is located, and the regional window where the truck is detected is labeled.

[0066] The expression of the classification decision function is as follows:

[0067] ;

[0068] In the formula, Indicates the presence of trucks in the region block, represents the sign function that converts the scores in brackets into discrete category labels. represents the Lagrange multiplier, represents the category of support vector, Indicates that the area block does not contain trucks. Indicates that the area block contains trucks. Represents the kernel function used to calculate the truck features in the area block With support vector The similarity between represents the bias term, Represents the total number of data in the training dataset.

[0069] What needs to be explained is that after processing the image data, the area where the truck is located is obtained, and these data are used as training data sets, support vector Refers to the key data points selected from the training dataset.

[0070] The calculation formula for the degree of overlap of each area window is as follows:

[0071] ;

[0072] In the formula, Display area window and area window The degree of overlap, Display area window and area window The area of ​​the intersection of Display area window and area window The area of ​​the union of .

[0073] The image data is divided into multiple blocks through a fully convolutional network. Each block may contain one or more objects. Considering that the size of the truck may vary depending on the distance from the camera in the capture device, a scanning window is configured according to the size of the truck. The blocks are scanned and identified through scanning windows of different sizes to capture targets of different sizes. At the same time, blocks that do not contain trucks, that is, the background, need to be excluded. Truck features are extracted in each block, including directional gradient histograms, etc. The sign of the classification decision function is used to determine whether there is a truck in the block. When When , it means there is a truck in the area block. When , it means that there is no truck in the area block. After the judgment is completed, multiple overlapping area windows and the confidence score of each area window will be obtained. The overlap degree of the two area windows is calculated, and the degree threshold is set. When the overlap degree of the two area windows is greater than the degree threshold, the area window with a lower confidence score is removed, and pairwise comparisons are continuously performed to finally obtain the area window where the truck is located and mark it with a label.

[0074] S4. Perform behavioral analysis on the detected trucks to identify the trucks' red light running behavior. If the trucks run red lights, the image data is marked as a violation image and saved in the database. Meanwhile, the privacy of the violation image is protected.

[0075] The detected trucks are analyzed through behavior analysis strategies to identify the trucks' red light running behaviors. The behavior analysis strategies are as follows: Figure 4 As shown, specifically including:

[0076] By obtaining the regional window of the truck in different image data, tracking the position of the truck, and calculating the speed of the truck, the state of the traffic light is determined. Combining the speed of the truck and the state of the traffic light, a comprehensive judgment is made on the truck's red light running behavior.

[0077] The speed of the truck is calculated as follows:

[0078] ;

[0079] In the formula, represents the speed of the truck, Indicates time The location of the truck in the corresponding image data, Indicates time The location of the truck in the corresponding image data, Indicates the time difference between different image data;

[0080] The strategies for comprehensively judging the red light running behavior of trucks include:

[0081] If the truck crosses the stop line when the traffic light is red and the speed of the truck remains unchanged, the truck is considered to have run a red light.

[0082] If the truck's position and speed indicate that it cannot stop before the traffic light turns red when the traffic light is green or yellow, the truck is considered to have run a red light.

[0083] If the truck's position and speed show that it stops before the traffic light turns red when the traffic light is green or yellow, then the truck does not run a red light.

[0084] After obtaining the regional window of the truck in different image data, the position of the truck is tracked by the target tracking algorithm, and the movement trajectory of the truck is established. The speed of the truck is estimated by calculating the position displacement of the truck in different image data divided by the time difference, and the image recognition technology is used to identify the state of the traffic light, including red light, yellow light or red light. The red light running behavior of the truck is comprehensively judged by combining the position, speed and the state of the traffic light at that time of the truck. When the traffic light is green or yellow, the position and speed of the truck are identified. If there is no trend of speed reduction, it is estimated whether the truck can stop before the traffic light turns red through the position and speed of the truck, so as to determine whether the truck has run a red light. When the traffic light is already red, the position of the truck exceeds the stop line and the speed remains unchanged and drives towards the traffic intersection. In this case, the truck has run a red light. The image data of the truck running a red light is marked as a violation image and saved in the database.

[0085] Non-violating vehicles and pedestrians in violation images are blurred to protect personal privacy. License plate numbers are desensitized, including replacing characters or blurring the license plate area to prevent personal information leakage. Stored and transmitted violation images are encrypted to ensure the security of violation images. Only the minimum amount of data required for violation evidence is stored, and no additional personal information is stored. Privacy protection measures are audited regularly to ensure compliance with the latest laws and regulations. At the same time, the retention period of violation images is defined and the violation images are securely destroyed after expiration.

[0086] S5. If an illegal image is detected, an alarm will be automatically triggered and an alarm message will be sent to relevant departments.

[0087] When an illegal image is detected, an alarm is automatically triggered and an alarm message is sent to relevant departments. The alarm message includes time, vehicle information and red light running location, where time refers to the specific timestamp when the illegal image occurs, vehicle information refers to detailed information of the truck that runs a red light, including license plate number, vehicle model and color, etc. The red light running location refers to the specific location where the truck runs a red light, including the name of the traffic intersection. The format of the alarm message is defined as: alarm message = {'time':'specific time in the illegal image', 'vehicle information':'truck vehicle details in the illegal image', 'red light running location':'location of the truck in the illegal image'}. The alarm message is automatically sent to relevant departments through a dedicated secure network protocol, and the confirmation of the recipient is recorded to ensure the accuracy and timeliness of information transmission.

[0088] Example 2

[0089] In this embodiment, the device for capturing a truck running a red light based on image recognition includes a camera for capturing image data of a truck passing through a traffic intersection.

[0090] The capture device uses a 2MP telephoto camera to capture license plates within 50 meters, and a 4MP wide-angle camera to detect vehicle behavior. At the same time, the capture device is equipped with high-precision GPS and BD dual-mode, which can achieve an error of less than 5cm, and accurately capture image data of trucks passing through traffic intersections.

[0091] Example 3

[0092] In this embodiment, a computer device is provided, including a memory and a processor, the memory is used to store instructions, and the processor is used to execute the instructions, so that the computer device executes the steps of implementing the above-mentioned method for capturing trucks running red lights based on image recognition.

[0093] Example 4

[0094] In this embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed, the steps of the above-mentioned method for capturing a truck running a red light based on image recognition are implemented.

[0095] The computer-readable storage medium includes: a U disk, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk, and other media for storing program codes.

[0096] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A method for capturing trucks running red lights based on image recognition, characterized in that: include: Capture image data of trucks passing through traffic intersections, and dynamically adjust the parameters of the capture device and the image capture frequency through image capture strategies; Performing image preprocessing on the captured image data; Perform target detection on trucks in the image data to determine the existence of trucks; Conduct behavioral analysis on detected trucks to identify red light running behaviors. If a truck runs a red light, the image data will be marked as a violation image and saved in the database. At the same time, the privacy of the violation image will be protected. If an illegal image is detected, an alarm will be automatically triggered.

2. The method for capturing a truck running a red light based on image recognition as claimed in claim 1, characterized in that: Capture image data of a truck passing through a traffic intersection, and dynamically adjust the parameters of the capture device and the image capture frequency through the image capture strategy, wherein the image capture strategy includes: According to the size and shape of the traffic intersection, the number and location of the capture devices are determined, and the height and angle are configured for each capture device. The resolution of the camera of each capture device is configured according to the requirements of truck identification, the frame rate is selected according to the storage requirement relationship, and the exposure time is automatically adjusted according to the lighting conditions. According to the real-time traffic flow and the preset time interval, the working state of the capture device is dynamically adjusted, and the traffic flow threshold is configured. If the real-time traffic flow is greater than the traffic flow threshold, the capture frequency of the capture device is increased; if the real-time traffic flow is equal to the traffic flow threshold, the capture device captures the image data at the preset time interval; if the real-time traffic flow is less than the traffic flow threshold, the capture frequency of the capture device is reduced.

3. The method for capturing a truck running a red light based on image recognition as claimed in claim 2, characterized in that: The truck in the image data is detected by a target detection strategy, and the target detection strategy includes: The image data is divided into regional blocks, and the scanning window is configured according to the size of the truck. The regional blocks are scanned and identified through the scanning window, and the regional blocks that do not contain the truck are excluded. At the same time, the truck features are extracted in each regional block. The existence of the truck in the regional block is judged by the classification decision function, and the overlapping regional windows are obtained. The overlapping degree of each regional window is calculated to obtain the regional window where the truck is located, and the regional window where the truck is detected is labeled.

4. The method for capturing a truck running a red light based on image recognition as claimed in claim 3, characterized in that: The calculation formula for the overlap degree of each regional window is as follows: ； In the formula, Display area window and area window The degree of overlap, Display area window and area window The area of ​​the intersection of Display area window and area window The area of ​​the union of .

5. The method for capturing a truck running a red light based on image recognition as claimed in claim 4, characterized in that: The detected truck is subjected to behavior analysis through a behavior analysis strategy to identify the truck's red light running behavior. The behavior analysis strategy includes: By obtaining the regional window of the truck in different image data, tracking the position of the truck, and calculating the speed of the truck, the state of the traffic light is determined. Combining the speed of the truck and the state of the traffic light, a comprehensive judgment is made on the truck's red light running behavior.

6. The method for capturing a truck running a red light based on image recognition as claimed in claim 5, characterized in that: The calculation formula of the speed of the truck is as follows: ； In the formula, represents the speed of the truck, Indicates time The location of the truck in the corresponding image data, Indicates time The location of the truck in the corresponding image data, Indicates the time difference between different image data.

7. The method for capturing a truck running a red light based on image recognition as claimed in claim 6, characterized in that: The comprehensive judgment of a truck's red light running behavior includes: If the truck crosses the stop line when the traffic light is red and the speed of the truck remains unchanged, the truck is considered to have run a red light. If the truck's position and speed indicate that it cannot stop before the traffic light turns red when the traffic light is green or yellow, the truck is considered to have run a red light. If the truck's position and speed show that it stops before the traffic light turns red when the traffic light is green or yellow, then the truck does not run a red light.

8. A truck red light running capture device based on image recognition, used to implement a truck red light running capture method based on image recognition as described in any one of claims 1 to 7, characterized in that: include: Cameras are used to capture image data of trucks passing through traffic intersections.

9. A computer device, characterized in that: include: A memory for storing instructions; A processor is used to execute the instructions so that the computer device executes the method for capturing a truck running a red light based on image recognition as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, the method for capturing a truck running a red light based on image recognition as described in any one of claims 1 to 7 is implemented.

Images