CN-122023775-A - Vehicle lane occupation operation detection method and device, electronic equipment and medium

CN122023775ACN 122023775 ACN122023775 ACN 122023775ACN-122023775-A

Abstract

The embodiment of the application discloses a vehicle lane occupation operation detection method, a device, electronic equipment and a storage medium, which are used for acquiring a road monitoring video, stopping inhibition identification information of a preset area and road function partition information, carrying out multi-dimensional target detection on the road monitoring video, extracting the stopping characteristics of a candidate vehicle, the behavior characteristics of the candidate vehicle and the morphological characteristics of a suspected commodity display area if the candidate vehicle stopped in the stopping area, the aggregation personnel surrounding the candidate vehicle and the suspected commodity display area exist in the detection result, and carrying out fusion analysis based on the stopping characteristics of the candidate vehicle, the behavior characteristics of the aggregation personnel and the morphological characteristics of the suspected commodity display area to determine whether the behavior of the vehicle lane occupation operation is implemented by using the candidate vehicle or not.

Inventors

CHEN LUBIN
CHEN Jin
LI LI

Assignees

深圳市小鹰视界智能有限公司

Dates

Publication Date: 20260512
Application Date: 20260202

Claims (10)

1. A vehicle lane occupancy detection method, comprising: acquiring a road monitoring video and forbidden stop identification information of a preset area and road function partition information; performing multi-dimensional target detection on the road monitoring video; If a candidate vehicle parked in a forbidden region, an aggregation person surrounding the candidate vehicle and a suspected commodity display region exist in the detection result, parking characteristics of the candidate vehicle, behavior characteristics of the aggregation person and morphological characteristics of the suspected commodity display region are extracted; And carrying out fusion analysis based on the parking characteristics of the candidate vehicles, the behavior characteristics of the aggregated personnel and the morphological characteristics of the suspected commodity display area, and determining whether the personnel corresponding to the candidate vehicles have the behavior of using the candidate vehicles to implement vehicle lane occupation operation.
2. The method for detecting vehicle lane occupation operation according to claim 1, wherein the multi-dimensional object detection of the road monitoring video comprises: Detecting vehicles in each frame of image in the road monitoring video, and screening candidate vehicles parked in the forbidden parking area by combining forbidden parking identification information and road function partition information of the preset area; detecting personnel in each frame of image in the road monitoring video, and counting the number of the personnel in the preset range of the candidate vehicle in each frame of image to obtain the aggregated personnel surrounding the candidate vehicle; performing region segmentation on each frame of image in the road monitoring video, and detecting a suspected commodity display region of which the periphery of the candidate vehicle accords with a preset commodity display form; And respectively recording the parking time of the candidate vehicle, the continuous gathering time of the gathering personnel and the existence time of the suspected commodity display area based on the time dimension of the road monitoring video.
3. The vehicle lane keeping operation detecting method according to claim 3, wherein the screening out the candidate vehicles parked in the parking area by combining the parking keeping identification information and the road function partition information of the preset area comprises: defining a forbidden region range in a frame image of the road monitoring video according to the forbidden identification information and the road function partition information; calculating the position relation between the detected coordinates of the central point of the vehicle and the range of the forbidden region; if the coordinates of the central point of the vehicle fall in the range of the forbidden region and the position of the vehicle is kept stable in the images of the continuous preset frames, the vehicle is determined to be a candidate vehicle parked in the forbidden region.
4. The vehicle lane-occupation operation detection method of claim 2, wherein the extracting the parking characteristics of the candidate vehicle, the behavior characteristics of the aggregated person, and the morphological characteristics of the suspected commodity display area comprises: The parking characteristics of the candidate vehicle comprise a difference value that the parking position deviates from the distance of a parking area of a clutch gauge and the parking time exceeds a preset compliance time; the behavior characteristics of the aggregated personnel comprise an aggregated personnel number peak value, a continuous aggregation duration, an average distance between the personnel and the candidate vehicle and a residence frequency of the personnel; the morphological characteristics of the suspected commodity display area comprise area of the area, regularity of outline of the area, connection relation between the area and the candidate vehicle and dynamic change amplitude of the area in time dimension.
5. The vehicle lane operation detection method according to claim 4, wherein the detecting the suspected goods presentation area in which the surroundings of the candidate vehicle meet a preset goods presentation form includes: Carrying out semantic segmentation on the image areas around the candidate vehicles to obtain a plurality of sub-areas to be detected; Calculating the area and the contour regularity of each sub-area to be detected and the edge distance between the sub-area and the candidate vehicle; And if the area of the sub-area to be detected is within the range of the preset commodity area, the contour regularity is greater than a preset regularity threshold, and the distance between the sub-area to be detected and the edge of the candidate vehicle is smaller than a preset distance threshold, determining the sub-area to be detected as a suspected commodity display area.
6. The method according to claim 4, wherein the determining whether the person corresponding to the candidate vehicle has a behavior of performing the vehicle lane operation using the candidate vehicle based on the fusion analysis of the parking characteristics of the candidate vehicle, the behavior characteristics of the aggregated person, and the morphological characteristics of the suspected merchandise display area includes: respectively setting weight coefficients for the parking characteristics of the candidate vehicles, the behavior characteristics of the aggregation personnel and the morphological characteristics of the suspected commodity display area; According to the weight coefficient, calculating a fusion characteristic value, wherein the calculation formula of the fusion characteristic value is as follows: fusion eigenvalue = α x dock feature + β x aggregate behavioral feature + γ x morphological feature; Wherein, alpha, beta and gamma are weight coefficients of parking characteristics, aggregation behavior characteristics and morphological characteristics respectively, and alpha+beta+gamma=1; If the fusion characteristic value is larger than a preset fusion threshold value, determining that the person corresponding to the candidate vehicle has a behavior of using the candidate vehicle to implement vehicle lane occupation operation, and if the fusion characteristic value is smaller than or equal to the preset fusion threshold value, determining that the behavior does not exist.
7. The vehicle lane departure management detection method of claim 6, wherein the method further comprises: according to different road scenes and time period characteristics, adjusting the values of alpha, beta and gamma and the preset fusion threshold; The method comprises the steps of increasing the value of beta in a daytime people stream intensive scene, increasing the value of gamma in a night light darker scene, and increasing the value of alpha in an urban arterial road scene.
8. A vehicle lane operation detection apparatus, comprising: The acquisition module is used for acquiring the road monitoring video, the forbidden stop identification information of the preset area and the road function partition information; The detection module is used for carrying out multi-dimensional target detection on the road monitoring video; the extraction module is used for extracting the parking characteristics of the candidate vehicles, the behavior characteristics of the aggregation personnel and the morphological characteristics of the suspected commodity display area if the candidate vehicles parked in the forbidden parking area, the aggregation personnel surrounding the candidate vehicles and the suspected commodity display area exist in the detection result; The determining module is used for carrying out fusion analysis based on the parking characteristics of the candidate vehicle, the behavior characteristics of the aggregated personnel and the morphological characteristics of the suspected commodity display area, and determining whether the personnel corresponding to the candidate vehicle have the behavior of using the candidate vehicle to implement vehicle lane occupation operation.
9. An electronic device comprising a memory, a processor and a processor program stored on the memory and executable on the processor, wherein the processor executes the program steps of the vehicle lane management method of any one of claims 1 to 5.
10. A storage medium storing a computer processing program capable of being loaded by a processor and executing the vehicle lane management detection method according to any one of claims 1 to 5.

Description

Vehicle lane occupation operation detection method and device, electronic equipment and medium Technical Field The application relates to the technical field of data analysis and processing, in particular to a vehicle lane occupation operation detection method, a device, electronic equipment and a storage medium. Background The vehicle occupying-road operation behavior is usually represented by that an operator drives a vehicle with a storage space such as a household car, SUV, MPV and the like, stops at a forbidden stop area such as an urban arterial road, a secondary arterial road, a sidewalk edge and the like, and the vehicle is used as a basis for carrying out commodity sales activities. The traffic behavior of the vehicle occupying the road not only seriously disturbs the urban road traffic order, but also forms a remarkable threat to public safety. On one hand, the management vehicles occupy the effective traffic space of the motor vehicle lanes or the sidewalks, so that the traffic capacity of the roads is reduced, traffic jam is easily caused, on the other hand, customers purchasing goods are often gathered around the vehicles, partially stand in the motor vehicle lanes even, and form a mixed traffic state with the passing vehicles, so that the occurrence risk of traffic accidents is greatly increased. In addition, the management activities generally lack standard management, so that the problems of noise pollution, garbage accumulation and the like are easy to generate, and the urban environment is destroyed. Compared with the traditional mobile stall, the vehicle road occupation operation has stronger concealment and mobility. The operator can utilize spaces such as a trunk, a roof luggage rack, a carriage and the like of the vehicle to rapidly display commodities, and can finish commodity storage and leave the scene in a short time when a law enforcement officer patrols, so that great challenges are brought to evidence collection and investigation of illegal behaviors. In the prior art, detection of vehicle lane occupation operation mainly depends on two modes, namely manual inspection, inspection in key areas is carried out regularly or irregularly by law enforcement personnel such as urban management, traffic police and the like, and on-site disposal is carried out after illegal behaviors are found. However, the manual inspection has obvious short plates, so that a large amount of human resources are required to be input, the detection efficiency is low, the detection is limited by the working time and inspection range of law enforcement personnel, all-weather and full-coverage monitoring is difficult to realize, and the illegal behaviors on night and remote road sections are more difficult to effectively manage and control. Secondly, an automatic detection technology based on video monitoring is focused on the interaction behavior of a trunk of a vehicle and personnel in most of the existing automatic detection schemes, and whether the operation behavior exists is judged by detecting the switching state of the trunk and the hand-stretching operation of the personnel. However, existing automated detection schemes have significant technical limitations. In an actual application scene, part of operators can adopt an operation mode of not frequently opening the trunk to avoid detection, for example, goods are placed on the ground around a vehicle and a folding goods shelf, or the goods are directly held by hands to be sold to passers-by, at the moment, illegal behaviors cannot be accurately identified only through trunk interaction information, in addition, for a scene of multi-person cooperation operation, part of operators can be responsible for soliciting customers, part of operators are responsible for taking and placing goods, and the existing scheme is difficult to comprehensively judge multi-dimensional illegal characteristics. The existence of the scenes leads to lower detection accuracy of the prior art, a large number of illegal behaviors can escape supervision, and the actual requirements of urban management can not be met. Therefore, a technical scheme for detecting the lane occupation operation of the vehicle by integrating multidimensional features is needed to break through the limitation of the prior art, and the precise identification of various lane occupation operation scenes of the vehicle is realized by integrating various information such as vehicle parking features, personnel gathering features, commodity display features and the like, so that the detection efficiency and accuracy are improved, and efficient and reliable technical support is provided for urban management departments. Disclosure of Invention The embodiment of the application provides a vehicle lane-occupying operation detection method, electronic equipment, a device and a storage medium, which aim to solve the problems of single detection dimension, low accuracy and limited applicable scene in the existing ve