A method for collecting and correcting evidence of large trucks running red lights based on road surveillance video

Abstract

The present invention relates to the field of image recognition technology, and specifically discloses a method for collecting and correcting evidence of large trucks running red lights based on road surveillance videos, comprising: acquiring road surveillance videos in real time; automatically adjusting the position of a virtual stop line when a large truck is seen in the road surveillance video, and sequentially capturing a first image of the large truck running, a second image of the large truck running, and a third image of the large truck running after the traffic light switches to red; correcting the first image of the large truck running and the second image of the large truck running when both the first image of the large truck running and the second image of the large truck running fail to meet the standards; and using the corrected first image of the large truck running and the corrected second image of the large truck running as the first image of the violation and the second image of the violation, respectively. The present invention can improve the success rate of collecting evidence of large trucks running red lights, curb large trucks running red lights by strengthening post-event penalties, and effectively reduce the incidence of large truck traffic accidents.

Description

Technical Field

[0001] The present invention relates to the field of image recognition technology, and more particularly to a method for collecting and correcting evidence of a large truck running a red light based on road monitoring videos. Background Art

[0002] Off-site evidence collection for large trucks running red lights follows standards such as the "Technical Specifications for Image Evidence Collection of Road Traffic Safety Violations" and the "General Technical Conditions for Automatic Red Light Running Recording Systems". Generally, at least three photos of the violation are required. When the traffic light in front is red, the front wheels of the vehicle cross the stop line and the first photo is taken. When the rear wheels cross the stop line again, another photo is taken. The third photo is taken when the vehicle reaches the zebra crossing or has crossed the zebra crossing.

[0003] The automatic red light running recording system has a high accuracy rate in capturing evidence photos for general small and medium-sized vehicles. However, for large trucks running red lights, the photos captured are not very accurate due to the following reasons.

[0004] First, when a large truck runs a red light, it is easy to trigger a photo taking action before the front wheels cross the stop line.

[0005] Existing electronic police systems generally trigger photography based on video. The specific method is to draw a virtual stop line along the actual stop line in the image, and trigger photography when part of the vehicle touches the virtual stop line.

[0006] Since the virtual stop line is set according to the height of most small and medium-sized vehicles, and large trucks are tall, it is easy to trigger the photo in advance when shooting from the rear. At this time, the front wheels of the large truck may not have crossed the stop line, resulting in insufficient evidence.

[0007] Second, the rear compartments of some large trucks are too high and too large, blocking the front wheels and making it impossible to determine whether the front wheels have crossed the line.

[0008] Some large fence-type trucks and vans carry bulky goods, and the electronic police at the intersection are not installed at a high height and the shooting angle is directly behind. The carriage and cargo in the image will block the front wheels of the truck. In addition, the shooting starts when the truck just crosses the stop line. It is difficult to determine whether the front wheels of the truck have crossed the line based on the photo alone.

[0009] Third, some large trucks have long bodies, which resulted in the rear wheels not crossing the stop line when the second photo was taken.

[0010] After the red light electronic police capture the first image, the first method is to delay the second image based on experience. The second method is to detect that there are no more vehicles at the virtual stop line and start the second image. The third method is to set a virtual starting line in front of the stop line and capture the second image as the vehicle passes. The first method is not effective because the delay is insufficient due to the length of the truck body, and the rear wheels have not yet passed the stop line. The second method, due to algorithmic issues, requires the image of the stop line area to remain unchanged for a period of time to determine that no vehicles have passed, which will cause delays in capturing the image. Moreover, if other vehicles are following too closely behind the truck, it may result in missed images. The third method is also due to the length of the truck body. When the front of the truck reaches the virtual starting line of the rear wheels, the rear wheels have not yet passed the stop line, resulting in the second image being captured prematurely.

[0011] For these reasons, the images of trucks running red lights collected by electronic police at off-site locations are prone to controversy, making it difficult to penalize trucks running red lights, reducing the deterrent effect on truck violations and increasing safety risks. Therefore, a method for correcting evidence collected by trucks running red lights is urgently needed. Summary of the Invention

[0012] In order to address the deficiencies in the prior art, the present invention provides a method for correcting evidence collection of large trucks running red lights based on road surveillance videos, which can improve the success rate of evidence collection of large trucks running red lights, curb the illegal behavior of large trucks running red lights by strengthening post-event penalties, and effectively reduce the incidence of large truck traffic accidents on national, provincial and urban roads.

[0013] As a first aspect of the present invention, a method for collecting and correcting evidence of a large truck running a red light based on road surveillance video is provided, comprising the following steps:

[0014] Step S1: Acquire road surveillance video of the intersection in real time;

[0015] Step S2: When a large truck is traveling in the road monitoring video, the position of the virtual stop line in the road monitoring video is automatically adjusted, and after the traffic light in front of the large truck switches to a red light, a first large truck driving image, a second large truck driving image, and a third large truck driving image are sequentially photographed; wherein, the first large truck driving image is photographed when the front wheels of the large truck pass through the virtual stop line, the second large truck driving image is photographed when the rear wheels of the large truck pass through the virtual stop line, and the third large truck driving image is photographed when the large truck reaches or passes the zebra crossing position, and the third large truck driving image is used as the third large truck driving image as evidence of the illegal red light running by the large truck;

[0016] Step S3: When both the first large truck driving image and the second large truck driving image are unqualified, the first large truck driving image and the second large truck driving image are corrected to obtain a corrected first large truck driving image and a corrected second large truck driving image;

[0017] Step S4: using the first corrected image of the truck driving as the first image of evidence of the truck running a red light violation, and using the second corrected image of the truck driving as the second image of evidence of the truck running a red light violation.

[0018] Furthermore, the steps S3 and S4 further include:

[0019] When both the first truck driving image and the second truck driving image are unqualified, identifying the driving track of the truck in the road monitoring video;

[0020] When the driving trajectory of the large truck in the road monitoring video intersects the virtual stop line, intercepting the current driving image of the large truck in the road monitoring video, and using the current driving image of the large truck in the road monitoring video as the first evidence image of the large truck running a red light violation, so as to achieve correction of the first driving image of the large truck;

[0021] Intercepting multiple frames of large truck driving images between the first large truck running a red light violation evidence image and the third large truck running a red light violation evidence image from the road surveillance video;

[0022] Selecting a current middle frame of the truck driving image from the plurality of frames of the truck driving image by a binary method; and identifying whether the rear wheels of the truck in the current middle frame of the truck driving image are located on the virtual stop line;

[0023] If the rear wheels of the truck in the current intermediate frame of the truck driving image are located on the virtual stop line, the current intermediate frame of the truck driving image is used as the second evidence image of the truck running a red light violation;

[0024] If the rear wheels of the truck in the current intermediate frame truck driving image are not located on the virtual stop line, the next intermediate frame truck driving image is selected backward or forward by binary selection, and this cycle is repeated until the rear wheels of the truck in the next intermediate frame truck driving image are located on the virtual stop line, and the next intermediate frame truck driving image is used as the second evidence image of the truck running a red light violation, so as to realize the correction of the second truck driving image.

[0025] Furthermore, if the rear wheels of the truck in the current intermediate frame of the truck driving image are not located on the virtual stop line, then the next intermediate frame of the truck driving image is selected by a backward or forward dichotomy method, further comprising:

[0026] If the rear wheels of the truck in the current intermediate truck driving image are in front of the virtual stop line, then the next intermediate truck driving image is selected using a backward binary search method, and it is identified whether the rear wheels of the truck in the next intermediate truck driving image are located on the virtual stop line, and this cycle is repeated until the rear wheels of the truck in the next intermediate truck driving image are located on the virtual stop line;

[0027] If the rear wheels of the truck in the current intermediate frame truck driving image are behind the virtual stop line, the next intermediate frame truck driving image is selected by forward dichotomy, and it is identified whether the rear wheels of the truck in the next intermediate frame truck driving image are located on the virtual stop line. This cycle is repeated until the rear wheels of the truck in the next intermediate frame truck driving image are located on the virtual stop line.

[0028] Furthermore, when both the first truck driving image and the second truck driving image are unqualified, identifying the truck driving trajectory in the road monitoring video further includes:

[0029] When both the first truck driving image and the second truck driving image are unqualified, a multi-target tracking algorithm is applied to track the truck in the road monitoring video, and the driving trajectory of the truck is identified in combination with a video target recognition algorithm.

[0030] The method for correcting evidence of large trucks running red lights based on road monitoring videos provided by the present invention has the following advantages: through detailed analysis of videos of large trucks running red lights, photos of evidence of red light running that meet the requirements are intercepted to correct the pictures of evidence of large trucks running red lights, thereby improving the success rate of evidence collection of large trucks running red lights, and curbing the illegal behavior of large trucks running red lights by strengthening post-event penalties, thereby effectively reducing the incidence of large truck traffic accidents on national, provincial and urban roads. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present invention, but do not constitute a limitation of the present invention.

[0032] Figure 1 The present invention provides a flowchart of a method for collecting and correcting evidence of a large truck running a red light based on road monitoring video.

[0033] Figure 2This is a flowchart of a specific implementation method of a method for collecting and correcting evidence of a large truck running a red light based on road monitoring video provided by the present invention. DETAILED DESCRIPTION

[0034] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments of the present invention may be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0035] To further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation method, structure, features and effects of a large truck red light running evidence correction method based on road monitoring video proposed by the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of the present invention.

[0036] It should be noted that the terms "first," "second," and the like in the specification and claims of the present invention and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate for the embodiments of the present invention described herein. In addition, the terms "including," "having," and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatuses.

[0037] In this embodiment, a method for collecting and correcting evidence of a large truck running a red light based on road monitoring video is provided. Figure 1 As shown, the method for collecting and correcting evidence of a large truck running a red light based on road monitoring video includes:

[0038] Step S1: Acquire road surveillance video of the intersection in real time;

[0039] It should be noted that the road surveillance video is an electronic police video, which is connected to the electronic police video through an interface module and undergoes preprocessing, such as image denoising and resolution adjustment.

[0040] Step S2: When a large truck is traveling in the road monitoring video, the position of the virtual stop line in the road monitoring video is automatically adjusted, and after the traffic light in front of the large truck switches to a red light, a first large truck driving image, a second large truck driving image, and a third large truck driving image are sequentially photographed; wherein, the first large truck driving image is photographed when the front wheels of the large truck pass through the virtual stop line, the second large truck driving image is photographed when the rear wheels of the large truck pass through the virtual stop line, and the third large truck driving image is photographed when the large truck reaches or passes the zebra crossing position, and the third large truck driving image is used as the third large truck driving image as evidence of the illegal red light running by the large truck;

[0041] It should be noted that the target recognition algorithm is embedded in the electronic police red light running capture algorithm, and the target recognition algorithm is used to identify in real time whether the vehicle approaching the virtual stop line in the road monitoring video is a large truck. When the vehicle approaching the virtual stop line in the road monitoring video is identified as a large truck, the position of the virtual stop line in the road monitoring video is automatically adjusted, thereby triggering the red light running capture at the appropriate virtual stop line position.

[0042] Step S3: When both the first large truck driving image and the second large truck driving image are unqualified, the first large truck driving image and the second large truck driving image are corrected to obtain a corrected first large truck driving image and a corrected second large truck driving image;

[0043] It should be noted that the off-site evidence image review algorithm that supports the identification of disputed evidence images is used to pre-review the first large truck driving image and the second large truck driving image. When the first large truck driving image and the second large truck driving image are both unqualified, the first large truck driving image and the second large truck driving image need to be corrected; when the first large truck driving image and the second large truck driving image are both qualified, the first large truck driving image and the second large truck driving image are used as the first large truck driving image and the second large truck driving image respectively as evidence images of the first large truck running a red light violation.

[0044] Step S4: using the first corrected image of the truck driving as the first image of evidence of the truck running a red light violation, and using the second corrected image of the truck driving as the second image of evidence of the truck running a red light violation.

[0045] Preferably, Figure 2 As shown, the steps S3 and S4 further include:

[0046] When both the first truck driving image and the second truck driving image are unqualified, identifying the driving track of the truck in the road monitoring video;

[0047] It should be noted that when both the first large truck driving image and the second large truck driving image are unqualified, the corresponding historical road surveillance video is retrieved from the electronic police for analysis.

[0048] Specifically, when both the first truck driving image and the second truck driving image are unqualified, identifying the truck driving trajectory in the road monitoring video further includes:

[0049] When both the first and second truck driving images are unqualified, a multi-target tracking algorithm (such as Deep SORT) is applied to track the truck in the road surveillance video, and the driving trajectory of the truck is identified in combination with a video target recognition algorithm.

[0050] When the driving trajectory of the large truck in the road monitoring video intersects the virtual stop line, intercepting the current driving image of the large truck in the road monitoring video, and using the current driving image of the large truck in the road monitoring video as the first evidence image of the large truck running a red light violation, so as to achieve correction of the first driving image of the large truck;

[0051] It should be noted that this method is suitable for situations where the electronic police have captured a truck running a red light, but the first evidentiary photo is disputed because it was taken too early. Because vehicle trajectory tracking is based on the vehicle's center point, when the truck's trajectory intersects the virtual stop line, it means that the middle of the vehicle has crossed the stop line. At this point, the truck's front wheels must have crossed the stop line, but the rear wheels have not yet passed the stop line. Therefore, the captured image at this time is more consistent with the requirements of the first evidentiary image of the truck running a red light, making it less likely to cause controversy.

[0052] In order to ensure the validity of the first image of evidence of the large truck running a red light violation, a red light running review algorithm may be used to review the first image of evidence of the large truck running a red light violation.

[0053] In case the second truck driving image is unqualified, multiple frames of truck driving images between the first truck running a red light violation evidence image and the third truck running a red light violation evidence image are intercepted from the road surveillance video according to the timestamp;

[0054] It should be noted that the image segment between the first image of the large truck showing illegal red light violation evidence and the third image of the large truck showing illegal red light violation evidence in the road surveillance video is framed second by second. For vehicles with faster speeds, frames are extracted at shorter intervals.

[0055] Selecting a current middle frame of the truck driving image from the plurality of frames of the truck driving image by a binary method; and identifying whether the rear wheels of the truck in the current middle frame of the truck driving image are located on the virtual stop line;

[0056] If the rear wheels of the truck in the current intermediate frame of the truck driving image are located on the virtual stop line, the current intermediate frame of the truck driving image is used as the second evidence image of the truck running a red light violation;

[0057] If the rear wheels of the truck in the current intermediate frame truck driving image are not located on the virtual stop line, the next intermediate frame truck driving image is selected backward or forward by binary selection, and this cycle is repeated until the rear wheels of the truck in the next intermediate frame truck driving image are located on the virtual stop line, and the next intermediate frame truck driving image is used as the second evidence image of the truck running a red light violation, so as to realize the correction of the second truck driving image.

[0058] It should be noted that the image of a truck's rear wheels passing the stop line has obvious features, such as the visible features of the truck's tail and the stop line. Therefore, the truck's rear wheel crossing line recognition algorithm can be trained by collecting sample images to identify whether the truck's rear wheels in the image are crossing the line.

[0059] Preferably, Figure 2 As shown, if the rear wheels of the truck in the current intermediate frame of the truck driving image are not located on the virtual stop line, then the next intermediate frame of the truck driving image is selected by a backward or forward dichotomy method, further comprising:

[0060] If the rear wheels of the truck in the current intermediate truck driving image are ahead of the virtual stop line (the rear wheels of the truck have not yet crossed the line), then the next intermediate truck driving image is selected using a backward binary search method, and it is identified whether the rear wheels of the truck in the next intermediate truck driving image are located on the virtual stop line. This cycle is repeated until the rear wheels of the truck in the next intermediate truck driving image are located on the virtual stop line.

[0061] If the rear wheels of the truck in the current intermediate frame truck driving image are behind the virtual stop line (the rear wheels of the truck have passed the line but are too far away from the stop line), the next intermediate frame truck driving image is selected by forward dichotomy, and it is identified whether the rear wheels of the truck in the next intermediate frame truck driving image are located on the virtual stop line. This cycle is repeated until the rear wheels of the truck in the next intermediate frame truck driving image are located on the virtual stop line.

[0062] The present invention provides a method for correcting evidence of large trucks running red lights based on road surveillance videos. The method analyzes the illegal videos of large trucks running red lights in detail, intercepts photos of evidence of red light running that meet the requirements, and corrects the pictures of evidence of large trucks running red lights, thereby improving the success rate of evidence collection of large trucks running red lights. By strengthening post-event penalties to curb the illegal behavior of large trucks running red lights, the incidence of large truck traffic accidents on national, provincial and urban roads is effectively reduced.

[0063] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment, it is not intended to limit the present invention. Any technician familiar with the present profession can make slight changes or modifications to equivalent embodiments using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for collecting and correcting evidence of a large truck running a red light based on road monitoring video, characterized in that: The method for collecting and correcting evidence of a large truck running a red light based on road monitoring video includes the following steps: Step S1: Acquire road surveillance video of the intersection in real time; Step S2: When a large truck is traveling in the road monitoring video, the position of the virtual stop line in the road monitoring video is automatically adjusted, and after the traffic light in front of the large truck switches to a red light, a first large truck driving image, a second large truck driving image, and a third large truck driving image are sequentially photographed; wherein, the first large truck driving image is photographed when the front wheels of the large truck pass through the virtual stop line, the second large truck driving image is photographed when the rear wheels of the large truck pass through the virtual stop line, and the third large truck driving image is photographed when the large truck reaches or passes the zebra crossing position, and the third large truck driving image is used as the third large truck driving image as evidence of the illegal red light running by the large truck; Step S3: When both the first truck driving image and the second truck driving image are unqualified, identifying the truck driving track in the road monitoring video; When the driving trajectory of the large truck in the road monitoring video intersects the virtual stop line, intercepting the current driving image of the large truck in the road monitoring video, and using the current driving image of the large truck in the road monitoring video as the first evidence image of the large truck running a red light violation, so as to achieve correction of the first driving image of the large truck; Intercepting multiple frames of large truck driving images between the first large truck running a red light violation evidence image and the third large truck running a red light violation evidence image from the road surveillance video; Selecting a current middle frame of the truck driving image from the plurality of frames of the truck driving image by a binary method; and identifying whether the rear wheels of the truck in the current middle frame of the truck driving image are located on the virtual stop line; If the rear wheels of the truck in the current intermediate frame of the truck driving image are located on the virtual stop line, the current intermediate frame of the truck driving image is used as the second evidence image of the truck running a red light violation; If the rear wheels of the truck in the current intermediate frame of the truck driving image are not located on the virtual stop line, the next intermediate frame of the truck driving image is selected by a backward or forward dichotomy method, and this cycle is repeated until the rear wheels of the truck in the next intermediate frame of the truck driving image are located on the virtual stop line, and the next intermediate frame of the truck driving image is used as the second evidence image of the truck running a red light violation, so as to achieve correction of the second truck driving image; Step S4: The first corrected image of the truck driving is used as the first image of evidence of the truck running a red light violation, and the second corrected image of the truck driving is used as the second image of evidence of the truck running a red light violation.

2. The method for collecting and correcting evidence of a large truck running a red light based on road monitoring video according to claim 1 is characterized in that: If the rear wheels of the truck in the current intermediate frame of the truck driving image are not located on the virtual stop line, then the next intermediate frame of the truck driving image is selected by a backward or forward dichotomy method, further comprising: If the rear wheels of the truck in the current intermediate truck driving image are in front of the virtual stop line, then the next intermediate truck driving image is selected using a backward binary search method, and it is identified whether the rear wheels of the truck in the next intermediate truck driving image are located on the virtual stop line, and this cycle is repeated until the rear wheels of the truck in the next intermediate truck driving image are located on the virtual stop line; If the rear wheels of the truck in the current intermediate frame truck driving image are behind the virtual stop line, the next intermediate frame truck driving image is selected by forward dichotomy, and it is identified whether the rear wheels of the truck in the next intermediate frame truck driving image are located on the virtual stop line. This cycle is repeated until the rear wheels of the truck in the next intermediate frame truck driving image are located on the virtual stop line.

3. The method for collecting and correcting evidence of a large truck running a red light based on road monitoring video according to claim 1 is characterized in that: When both the first truck driving image and the second truck driving image are unqualified, identifying the truck driving track in the road monitoring video further includes: When both the first truck driving image and the second truck driving image are unqualified, a multi-target tracking algorithm is applied to track the truck in the road monitoring video, and the driving trajectory of the truck is identified in combination with a video target recognition algorithm.

Images