CN110570664B - Automatic detection system for highway traffic incident - Google Patents

Abstract

The invention discloses an automatic detection system for expressway traffic incidents. The system includes: a data collection unit including a millimeter-wave radar and video collection equipment for collecting vehicle motion tracks and vehicle information; a data analysis and event discrimination unit for The vehicle trajectory and vehicle information collected by the acquisition unit determine whether the vehicle state is an abnormal driving state; the on-site alarm unit is used to send an alarm to the event information on the expressway site when the judgment result obtained by the data analysis and event judgment unit is abnormal driving state , and send an alarm to the driver; the background management unit is used to coordinate and control the data acquisition unit, data analysis and event identification unit and on-site alarm unit, and display the collected data information, identification results and alarm information. The present invention utilizes the fusion of millimeter-wave radar and video to improve the event detection range, make the event detection types more comprehensive, and realize all-weather automatic detection and early warning of expressway events.

Description

An automatic detection system for highway traffic incidents

technical field

The invention relates to the technical field of road traffic state detection and management, in particular to an automatic detection system for expressway traffic events based on millimeter-wave radar and video fusion.

Background technique

With the rapid development of the national economy, my country has formed a network of expressways extending in all directions, and the total mileage of expressways has increased year by year. Due to the characteristics of large flow, fast speed, and various types of vehicles in expressway traffic, expressway accidents often cause large losses and high casualty rates. Expressway accidents have become the focus of traffic management. In expressway accidents, the driver's uncivilized driving behavior is the main cause of the accident, mainly including overspeed, low speed, retrograde, illegal parking, illegal occupation of emergency lanes, frequent lane changes, etc. Timely and accurate detection of these abnormal driving behaviors, reminders to drivers, and uploading to the traffic management department for law enforcement and supervision can effectively reduce the probability of accidents and improve the management level of expressways.

The existing automatic monitoring of highway traffic incidents is mainly realized by video equipment, including fixed-point detection equipment and high-point video monitoring equipment. The fixed-point detection mainly adopts the combination of microwave radar or ground sense coil and video equipment, uses microwave radar or ground sense coil to detect the speed of the vehicle, and correlates with video equipment for illegal evidence collection. Traditional microwave radar cannot detect low-speed and stationary targets, and the detection range is small, and the number of detection targets is small. By deploying speed measurement and video equipment for each lane, only a few types of events such as overspeed, low speed, and emergency lane occupancy can be automatically monitored, and it is only suitable for small-scale cross-sectional traffic incident collection.

Although high-point video surveillance equipment can obtain comprehensive traffic incident information, it is mainly realized through manual inspection. Expressway event management is realized by managers manually judging the illegal parking and retrograde status of vehicles within the video range. This method consumes a lot of manpower and cannot realize all-weather automatic event monitoring. In addition, the accuracy and reliability of event collection will be seriously affected in scenes with poor lighting such as heavy fog, rainy days, backlighting, and directing light.

Contents of the invention

The purpose of the present invention is to provide an automatic detection system for expressway traffic incidents, so as to automatically monitor expressway traffic incidents more comprehensively and accurately.

To achieve the above object, the present invention provides an automatic detection system for expressway traffic incidents, the detection system comprising:

A data collection unit, including millimeter wave radar and video collection equipment, used to collect vehicle trajectory and vehicle information, the vehicle information including license plate and body features;

The data analysis and event discrimination unit is used to judge whether the vehicle state belongs to an abnormal driving state according to the vehicle movement trajectory and vehicle information collected by the data collection unit, and the abnormal driving state includes overspeed, low speed, illegal parking, retrograde, and emergency lane occupancy or frequent lane changes;

The on-site alarm unit is used to warn the event information on the expressway site and issue an alarm to the driver when the discrimination result obtained by the data analysis and event discrimination unit is an abnormal driving state;

The background management unit is used to coordinate and control the data collection unit, the data analysis and event judgment unit and the on-site alarm unit, and display the collected data information, judgment results and alarm information.

Optionally, the data analysis and event discrimination unit includes:

The vehicle motion track data preprocessing module is used to filter valid vehicle motion track data according to a preset threshold range; the preset threshold range includes a distance threshold, an angle threshold, a speed threshold and an RCS threshold;

The vehicle target trajectory extraction module is used to obtain two video frames at a time as two sampling points, and calculates the similarity of the vehicle target in the two sampling points according to the similarity principle, and determines that the same vehicle target has the largest similarity, according to The same vehicle target obtained multiple times forms a vehicle trajectory;

The vehicle state discrimination module is used for, according to the vehicle trajectory, for the vehicle state in the same vehicle target trajectory according to the maximum speed limit threshold, the minimum speed limit threshold, the speed is zero, the speed is negative, the emergency lane angle range and the driving lane angle Judging the state of the vehicle respectively to obtain the judgment result, the judgment result includes normal driving state and abnormal driving state, and the abnormal driving state includes overspeed, low speed, illegal parking, retrograde, occupying emergency lane or frequent lane change;

The vehicle feature extraction module is used to match the image collected by the millimeter-wave radar with the image collected by the video collection device, unify the coordinates, identify vehicle feature information in the image collected by the millimeter-wave radar, and The vehicle feature information is stored and uploaded in association with the vehicle trajectory;

The video forensics module is used to record and store the video stream containing the characteristic information of the vehicle.

Optionally, the vehicle trajectory data preprocessing module specifically includes:

_{The distance threshold setting sub-module is used to calculate the maximum detection distance and the minimum detection distance range [d min ,} d _max ], rejecting the detection distance data that is not within the range of the distance threshold; where the maximum detection distance and the minimum detection distance are calculated as follows:

The angle threshold setting submodule is used to reject the detection angle data not within the radar horizontal angle detection range according to the radar horizontal angle detection range [-σ _max , σ _max ]; wherein, σ _max is the radar horizontal maximum detection angle;

The speed threshold setting sub-module is used to set the speed detection range [v _min , v _max ] according to the driving characteristics of the vehicle, and reject the detected speed data not within the speed detection range; wherein, v _min is the maximum speed of the vehicle when it is traveling in reverse , is a negative value, and v _max is 160% times the expressway speed limit;

The RCS threshold setting sub-module is used to determine the RCS distribution range [r _min , r _max ] of the vehicle target, and eliminate the detected RCS data within the RCS distribution range, where r _min and r _max are the RCS statistical values of the vehicle target minimum and maximum values.

Optionally, the vehicle target track extraction module specifically includes:

计算第p帧采样点数据在第q帧时刻的预测值，所述第p帧和第q帧这两个采样点为连续帧或设定间隔的两帧，所述第p帧和第q帧分别包含M个车辆目标点和N个车辆目标点；The predicted value calculation sub-module is used to use the formula

Calculate the predicted value of the sampling point data of the pth frame at the time of the qth frame, the two sampling points of the pth frame and the qth frame are continuous frames or two frames with a set interval, the pth frame and the qth frame Contains M vehicle target points and N vehicle target points respectively;

计算不同采样帧中任意两目标点之间相异度；其中，1≤m≤M,1≤n≤N，

为无量纲化后第p帧采样点数据在第q帧时刻的预测值；

为无量纲化后第q帧时刻的实际检测值，μ_d,μ_v,μ_a,μ_r为权重系统，且满足μ_d+μ_v+μ_a+μ_r＝1；The dissimilarity calculation sub-module is used to use the formula

Calculate the dissimilarity between any two target points in different sampling frames; where, 1≤m≤M, 1≤n≤N,

is the predicted value of the sampling point data of the pth frame at the time of the qth frame after dimensionless;

is the actual detection value at the qth frame after dimensionless, μ _d , μ _v , μ _a , μ _r are the weight system, and satisfy μ _d + μ _v + μ _a + μ _r =1;

The maximum dissimilarity calculation submodule is used to calculate the maximum dissimilarity of all sampling frame samples in the historical time period T according to the formula max dif=max dif (m, n);

The similarity calculation submodule is used to calculate the similarity according to the formula sim (m, n)=max dif-dif (m, n);

The trajectory generation sub-module is used to select the objects with the highest similarity in the p-th frame and the q-th frame according to the similarity and classify them into one category to form the trajectory of the same vehicle.

Optionally, the vehicle state discrimination module specifically includes:

The speeding discrimination sub-module is used to judge the vehicle target whose speed exceeds the maximum speed limit threshold in the same vehicle target trajectory as speeding;

The low-speed discrimination sub-module is used to determine a vehicle target whose speed is lower than the minimum speed limit threshold in the same vehicle target trajectory as a low speed;

Violation of parking discrimination sub-module, used in the same vehicle target track, the existence of a vehicle target whose speed is zero is determined as a violation of parking;

The retrograde discrimination sub-module is used to judge the vehicle target whose speed is negative in the same vehicle target trajectory as retrograde;

Illegal occupancy of the emergency lane identification sub-module, used to determine the illegal occupancy of the emergency lane when the vehicle target angle is within the angle range of the emergency lane in the same vehicle target trajectory. The angle range of the emergency lane needs to be determined according to the installation angle of the millimeter wave radar equipment calibration;

The frequent lane change discrimination sub-module is used to judge the frequent lane change when the vehicle target angle is continuously changed to different lane angle ranges within the detection range in the same vehicle target trajectory, and the lane angle range needs to be determined according to Millimeter wave radar equipment installation angle calibration.

Optionally, the vehicle feature extraction module specifically includes:

The image fusion sub-module is used to match the images collected by the millimeter-wave radar and the images collected by the video, and unify the coordinate system;

The projection sub-module is used to detect the vehicle target information on the millimeter wave radar, project it in the image collected by the video, and identify the interest point area, and the interest point area represents the position of the vehicle target;

The feature recognition sub-module is used to define the vehicle target extraction boundary according to the point of interest area, and identify the vehicle features within the extraction boundary, including license plate identity features, body features, vehicle type and body color;

The correlation matching sub-module is used to correlate and match the vehicle feature information used for video image recognition with the trajectory information recognized by the millimeter-wave radar, store it in the processing unit, and upload it to the background management unit in real time.

Optionally, the background management unit includes a data display platform, a system operation and maintenance management platform, and an abnormal event review management platform.

Optionally, the data display platform is used to combine big data visualization and GIS applications to summarize the system, monitor point distribution, real-time traffic event reporting, traffic event statistics, traffic flow status, video surveillance status, and radar detection targets Track display status for dynamic display.

Optionally, the system operation and maintenance management platform is used for online management of intersection radar and video detection equipment, processing unit equipment, broadcast equipment, display equipment, and network equipment, including equipment basic information maintenance, equipment failure alarm or personnel authority manage.

Optionally, the abnormal event review management platform is used to review and enforce traffic abnormal events collected on-site, and perform manual review according to event types reported by millimeter-wave radar and video evidence collection information collected by video equipment to identify illegal vehicles , type of violation, time and penalty standard, and transmit the obtained information to the traffic control platform for law enforcement.

According to the specific embodiments provided by the invention, the invention discloses the following technical effects:

1. The present invention adopts a highway incident detection scheme with millimeter-wave radar as the main component and video as the supplementary component, so that the types of detected events are more comprehensive and accurate. In addition, the present invention can reduce the dependence of traditional video methods on light conditions, and can still realize event discrimination and on-site alarm prompts even under poor light conditions, so as to achieve the purpose of traffic management;

2. The present invention combines millimeter-wave radar detection targets with video images, which can reduce the amount of video analysis data processing and improve the efficiency of vehicle feature extraction. In addition, the event-based video recording acquisition method can also greatly reduce unnecessary data transmission and save network bandwidth;

3. The present invention not only proposes an automatic detection method for expressway incidents in terms of key core technologies, but also provides a systematic solution for high-speed traffic management. Through the combination of front-end warning prompts and background law enforcement management, the probability of accidents is reduced and the accident rate is improved. management efficiency.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required in the embodiments. Obviously, the accompanying drawings in the following description are only some of the present invention. Embodiments, for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without any creative effort.

FIG. 1 is a system block diagram of an automatic detection system for expressway traffic incidents provided by an embodiment of the present invention.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

The purpose of the present invention is to provide an automatic detection system for expressway traffic incidents, so as to automatically monitor expressway traffic incidents more comprehensively and accurately.

In order to make the above objects, features and advantages of the present invention more comprehensible, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

As shown in FIG. 1 , the automatic detection system for expressway traffic incidents provided by this embodiment includes a data acquisition unit 1 , a data analysis and event discrimination unit 2 , an on-site alarm unit 3 and a background management unit 4 .

Wherein, the data collection unit 1 includes a millimeter wave radar and video collection equipment, and the data collection unit 1 is used to collect vehicle trajectory and vehicle information, and the vehicle information includes license plate and vehicle body features.

The working frequency band of millimeter wave radar is between light wave and centimeter wave, and has excellent detection performance. In particular, FMCW adjustable frequency continuous millimeter wave radar has the advantages of high speed measurement accuracy, multi-target detection, high resolution and imaging capabilities, wide detection range, strong anti-interference ability, etc., and it also has good detection of low-speed moving objects and stationary objects Performance, to better make up for the shortcomings of traditional microwave radar. This embodiment adopts the combination of millimeter-wave radar and video acquisition, mainly uses millimeter-wave radar to track vehicle trajectory for event discrimination, and video acquisition is supplemented, and proposes an all-weather highway event detection and early warning system.

In practical applications, the millimeter-wave radar and video equipment are installed on the roadside poles. The millimeter-wave radar mainly detects the movement trajectory characteristics of vehicle targets, and the direction of the radar antenna is consistent with the direction of traffic flow. The millimeter-wave radar used can realize continuous trajectory tracking of multiple vehicle targets in 3-4 lanes within 150m at the same time, and the tracking frame rate is 50ms/time. Among them, in the k-th frame, the tracking information of the i-th vehicle target is as follows :

＜d _k (i), v _k (i), a _k (i), r _k (i)＞

in,

d _k (i) represents the straight-line distance between target i and radar, in m;

v _k (i) represents the driving speed of target i, unit m/s;

a _k (i) represents the azimuth angle of target i;

r _k (i) represents the RCS return energy value intensity of target i, in dB.

The video equipment mainly detects the license plate and body characteristics of the vehicle, and at the same time conducts video recording and evidence collection of the abnormal events judged to provide a basis for traffic law enforcement management.

The data analysis and event discrimination unit 2 is used to judge whether the vehicle state belongs to an abnormal driving state according to the vehicle trajectory and vehicle information collected by the data collection unit, and the abnormal driving state includes overspeed, low speed, illegal parking, retrograde, and emergency lane occupation or frequent lane changes;

The data analysis and event discrimination unit 2 is mainly executed by a processing unit deployed on the roadside, and the processing unit may be an industrial computer, a server, an embedded processor or other computing devices that meet data processing performance. The processing unit receives the real-time data transmitted from the millimeter-wave radar and video equipment, and performs the following steps: by cleaning the millimeter-wave radar data, eliminating background noise points, and obtaining a set of vehicle movement trajectory points; classifying the vehicle movement trajectory points, and extracting a single Vehicle trajectory data; according to the trajectory of a single vehicle, identify the types of events such as overspeed, low speed, frequent lane changes, emergency lane occupancy, illegal parking, and reverse traffic; use video equipment to collect road data and compare them with millimeter wave data Fusion to realize vehicle feature extraction, mainly including license plate number, vehicle type, body color, etc.; finally, through the video recording evidence collection sub-module, when a traffic incident occurs, the video data within the corresponding time period is stored as the basis for law enforcement.

Specifically, the data analysis and event discrimination unit 2 includes a vehicle trajectory data preprocessing module, a vehicle target trajectory extraction module, a vehicle state discrimination module, a vehicle feature extraction module and a video evidence collection module.

The data collected by the millimeter-wave radar contains a large amount of noise information, which mainly comes from the radar equipment itself and the road environment, such as sidelobe interference, false alarm noise, and metal objects such as road guardrails interfering with targets. It is necessary to filter the effective vehicle trajectory data by threshold analysis and elimination of the above noise. The vehicle motion track data preprocessing module is used to filter valid vehicle motion track data according to a preset threshold range; the preset threshold range includes a distance threshold, an angle threshold, a speed threshold and an RCS threshold. The vehicle motion trajectory data preprocessing module specifically includes:

_{The distance threshold setting sub-module is used to calculate the maximum detection distance and the minimum detection distance range [d min ,} d _max ], rejecting the detection distance data that is not within the range of the distance threshold; where the maximum detection distance and the minimum detection distance are calculated as follows:

The angle threshold setting submodule is used to reject the detection angle data not within the radar horizontal angle detection range according to the radar horizontal angle detection range [-σ _max , σ _max ]; wherein, σ _max is the radar horizontal maximum detection angle;

The speed threshold setting sub-module is used to set the speed detection range [v _min , v _max ] according to the driving characteristics of the vehicle, and reject the detected speed data not within the speed detection range; wherein, v _min is the maximum speed of the vehicle when it is traveling in reverse , is a negative value, and v _max is 160% times the expressway speed limit;

The RCS threshold setting sub-module is used to determine the RCS distribution range [r _min , r _max ] of the vehicle target, and eliminate the detected RCS data within the RCS distribution range, where r _min and r _max are the RCS statistical values of the vehicle target minimum and maximum values.

The vehicle target trajectory extraction module is used to obtain two video frames each time as two sampling points, calculate the similarity of the vehicle target in the two sampling points according to the similarity principle, and determine that the same vehicle target has the largest similarity, according to The same vehicle target obtained multiple times forms a vehicle trajectory; the vehicle target trajectory extraction module specifically includes:

计算第p帧采样点数据在第q帧时刻的预测值，所述第p帧和第q帧这两个采样点为连续帧或设定间隔的两帧，所述第p帧和第q帧分别包含M个车辆目标点和N个车辆目标点；The predicted value calculation sub-module is used to use the formula

Calculate the predicted value of the sampling point data of the pth frame at the time of the qth frame, the two sampling points of the pth frame and the qth frame are continuous frames or two frames with a set interval, the pth frame and the qth frame Contains M vehicle target points and N vehicle target points respectively;

计算不同采样帧中任意两目标点之间相异度；其中，1≤m≤M,1≤n≤N，

为无量纲化后第p帧采样点数据在第q帧时刻的预测值；

为无量纲化后第q帧时刻的实际检测值，μ_d,μ_v,μ_a,μ_r为权重系统，且满足μ_d+μ_v+μ_a+μ_r＝1；The dissimilarity calculation sub-module is used to use the formula

Calculate the dissimilarity between any two target points in different sampling frames; where, 1≤m≤M, 1≤n≤N,

is the predicted value of the sampling point data of the pth frame at the time of the qth frame after dimensionless;

is the actual detection value at the qth frame after dimensionless, μ _d , μ _v , μ _a , μ _r are the weight system, and satisfy μ _d + μ _v + μ _a + μ _r =1;

The maximum dissimilarity calculation submodule is used to calculate the maximum dissimilarity of all sampling frame samples in the historical time period T according to the formula max dif=max dif (m, n);

The similarity calculation submodule is used to calculate the similarity according to the formula sim (m, n)=max dif-dif (m, n);

The trajectory generation sub-module is used to select the objects with the highest similarity in the p-th frame and the q-th frame according to the similarity and classify them into one category to form the trajectory of the same vehicle.

The vehicle state discrimination module is used for, according to the vehicle trajectory, for the vehicle state in the same vehicle target trajectory according to the maximum speed limit threshold, the minimum speed limit threshold, the speed is zero, the speed is negative, the emergency lane angle range and the driving lane angle Scope judges the state of the vehicle respectively, and obtains the judgment result, the judgment result includes normal driving state and abnormal driving state, and the abnormal driving state includes overspeed, low speed, illegal parking, reverse driving, occupying emergency lane or frequent lane change; the vehicle state judgment Modules specifically include:

The speeding discrimination sub-module is used to judge the vehicle target whose speed exceeds the maximum speed limit threshold in the same vehicle target trajectory as speeding;

The low-speed discrimination sub-module is used to determine a vehicle target whose speed is lower than the minimum speed limit threshold in the same vehicle target trajectory as a low speed;

Violation of parking discrimination sub-module, used in the same vehicle target track, the existence of a vehicle target whose speed is zero is determined as a violation of parking;

The retrograde discrimination sub-module is used to judge the vehicle target whose speed is negative in the same vehicle target trajectory as retrograde;

Illegal occupancy of the emergency lane identification sub-module, used to determine the illegal occupancy of the emergency lane when the vehicle target angle is within the angle range of the emergency lane in the same vehicle target trajectory. The angle range of the emergency lane needs to be determined according to the installation angle of the millimeter wave radar equipment calibration;

The frequent lane change discrimination sub-module is used to judge the frequent lane change when the vehicle target angle is continuously changed to different lane angle ranges within the detection range in the same vehicle target trajectory, and the lane angle range needs to be determined according to Millimeter wave radar equipment installation angle calibration.

The vehicle feature extraction module is used to match the image collected by the millimeter-wave radar with the image collected by the video acquisition device, unify the coordinates, identify vehicle feature information in the image collected by the millimeter-wave radar, and The vehicle feature information is stored and uploaded in association with the vehicle track; the vehicle feature extraction module specifically includes:

The image fusion sub-module is used to match the images collected by the millimeter-wave radar and the images collected by the video, and unify the coordinate system;

The projection sub-module is used to detect the vehicle target information on the millimeter wave radar, project it in the image collected by the video, and identify the interest point area, and the interest point area represents the position of the vehicle target;

The feature recognition sub-module is used to define the vehicle target extraction boundary according to the point of interest area, and identify the vehicle features within the extraction boundary, including license plate identity features, body features, vehicle type and body color;

The correlation matching sub-module is used to correlate and match the vehicle feature information used for video image recognition with the trajectory information recognized by the millimeter-wave radar, store it in the processing unit, and upload it to the background management unit in real time.

The video forensics module is used to record and store the video stream containing the characteristic information of the vehicle. When the millimeter-wave radar detects an abnormal event, it combines the video fusion method to extract the vehicle features, and at the same time, records the on-site video stream data, stores the recorded video stream data locally and uploads it to the background, providing a basis for law enforcement management. The present invention proposes that the video data recording and storage function is triggered only in real time when an abnormal event occurs on the road, thereby effectively saving network bandwidth resources, and especially reducing unnecessary data uploading for road sections with small traffic flow.

The on-site warning unit 3 is used for warning the event information on the expressway site and sending a warning to the driver when the judgment result obtained by the data analysis and event judging unit is an abnormal driving state. When an abnormal event occurs in the radar coverage area, an event information alarm will be sent to the highway scene to remind the driver to pay attention to safe driving. There are two main alarm methods: broadcast method, through the installation of loudspeaker equipment on the roadside, to implement sound reminders to drivers; induction screen reminder method: through the installation of LED induction screens and other display devices on the roadside, vehicle information and illegal information are monitored. Display to remind drivers to drive safely.

The background management unit 4 is used to coordinate and control the data collection unit, the data analysis and event judgment unit and the on-site alarm unit, and display the collected data information, judgment results and alarm information. The background management unit 4 includes a data display platform, a system operation and maintenance management platform, and an abnormal event review management platform.

The data display platform is used to combine big data visualization and GIS applications to dynamically perform system overview, monitoring point distribution, real-time traffic event reporting, traffic event statistics, traffic flow status, video monitoring status, and radar detection target track display status. exhibit.

The system operation and maintenance management platform is used for online management of intersection radar and video detection equipment, processing unit equipment, broadcast equipment, display equipment, and network equipment, including equipment basic information maintenance, equipment failure alarm or personnel authority management.

The abnormal event review and management platform is used to review and enforce the traffic abnormal events collected on the spot. According to the event type reported by the millimeter wave radar and the video evidence collection information collected by the video equipment, manual review is performed to clarify illegal vehicles, illegal types, time and penalty standards, and transmit the obtained information to the traffic control platform for law enforcement

In this embodiment, the fusion of millimeter wave radar and video is used to improve the event detection range, and the event detection types are more comprehensive, so as to realize all-weather automatic detection and early warning of expressway events. It solves the shortcomings of traditional microwave radar that cannot identify stationary vehicles and detect vehicle trajectory information, and the single video acquisition method is affected by light, which leads to the shortcomings of the existing highway incident detection system with few types of detection events and small range.

In this paper, specific examples have been used to illustrate the principle and implementation of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; meanwhile, for those of ordinary skill in the art, according to the present invention Thoughts, there will be changes in specific implementation methods and application ranges. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims (7)

1. A highway traffic incident automatic detection system, is characterized in that, described detection system comprises: A data collection unit, including millimeter wave radar and video collection equipment, used to collect vehicle trajectory and vehicle information, the vehicle information including license plate and body features; The data analysis and event discrimination unit is used to judge whether the vehicle state belongs to an abnormal driving state according to the vehicle movement trajectory and vehicle information collected by the data collection unit, and the abnormal driving state includes overspeed, low speed, illegal parking, retrograde, and emergency lane occupancy or frequent lane changes; The on-site alarm unit is used to warn the event information on the expressway site and issue an alarm to the driver when the discrimination result obtained by the data analysis and event discrimination unit is an abnormal driving state; The background management unit is used to coordinate and control the data collection unit, the data analysis and event judgment unit and the on-site alarm unit, and display the collected data information, judgment results and alarm information; The vehicle motion track data preprocessing module is used to filter valid vehicle motion track data according to a preset threshold range; the preset threshold range includes a distance threshold, an angle threshold, a speed threshold and an RCS threshold; the RCS threshold represents a radar cross-sectional area threshold; The vehicle target trajectory extraction module is used to obtain two video frames at a time as two sampling points, and calculates the similarity of the vehicle target in the two sampling points according to the similarity principle, and determines that the same vehicle target has the largest similarity, according to The same vehicle target obtained multiple times forms a vehicle trajectory; The vehicle state discrimination module is used for, according to the vehicle trajectory, for the vehicle state in the same vehicle target trajectory according to the maximum speed limit threshold, the minimum speed limit threshold, the speed is zero, the speed is negative, the emergency lane angle range and the driving lane angle Judging the state of the vehicle respectively to obtain the judgment result, the judgment result includes normal driving state and abnormal driving state, and the abnormal driving state includes overspeed, low speed, illegal parking, retrograde, occupying emergency lane or frequent lane change; The vehicle feature extraction module is used to match the image collected by the millimeter-wave radar with the image collected by the video collection device, unify the coordinates, identify vehicle feature information in the image collected by the millimeter-wave radar, and The vehicle feature information is stored and uploaded in association with the vehicle trajectory; The video evidence collection module is used to record and store the video stream containing the characteristic information of the vehicle; The vehicle motion trajectory data preprocessing module specifically includes: _{The distance threshold setting sub-module is used to calculate the maximum detection distance and the minimum detection distance range [d min ,} d _max ], rejecting the detection distance data that is not within the range of the distance threshold; where the maximum detection distance and the minimum detection distance are calculated as follows:

The angle threshold setting submodule is used to reject the detection angle data not within the radar horizontal angle detection range according to the radar horizontal angle detection range [-σ _max , σ _max ]; wherein, σ _max is the radar horizontal maximum detection angle; The speed threshold setting sub-module is used to set the speed detection range [v _min , v _max ] according to the driving characteristics of the vehicle, and reject the detected speed data not within the speed detection range; wherein, v _min is the maximum speed of the vehicle when it is traveling in reverse , is a negative value, and v _max is 160% times the expressway speed limit; The RCS threshold setting sub-module is used to determine the RCS distribution range [r _min , r _max ] of the vehicle target, and eliminate the detected RCS data within the RCS distribution range, where r _min and r _max are the RCS statistical values of the vehicle target The minimum value and the maximum value; the RCS distribution range represents the radar cross-sectional area distribution range, and the RCS data represents the radar cross-sectional area data; the vehicle target RCS statistical value represents the vehicle target radar cross-sectional area statistical value; The vehicle target trajectory extraction module specifically includes: The predicted value calculation sub-module is used to use the formula

Calculate the predicted value of the sampling point data of the pth frame at the time of the qth frame, the two sampling points of the pth frame and the qth frame are continuous frames or two frames with a set interval, the pth frame and the qth frame Contains M vehicle target points and N vehicle target points respectively; t _p , v _p , a _p , r _p are the time, driving speed, azimuth angle and RCS return energy value intensity in the sampling point data of the pth frame respectively; t _q is the moment of the qth frame; d' _q , v' _q , a' _q , r' _q is the predicted value of the sampling point data of the pth frame at the qth frame; the RCS returns the energy value intensity to represent the radar scattering The cross-sectional area returns the energy value intensity; The dissimilarity calculation sub-module is used to use the formula

Calculate the dissimilarity between any two target points in different sampling frames; where, 1≤m≤M, 1≤n≤N,

is the predicted value of the sampling point data of the pth frame at the time of the qth frame after dimensionless;

is the actual detection value at the qth frame after dimensionless, μ _d , μ _v , μ _a , μ _r are the weight system, and satisfy μ _d + μ _v + μ _a + μ _r =1; The maximum dissimilarity calculation submodule is used to calculate the maximum dissimilarity of all sampling frame samples in the historical time period T according to the formula maxdif=maxdif(m, n); The similarity calculation submodule is used to calculate the similarity according to the formula sim(m,n)=maxdif-dif(m,n); The trajectory generation sub-module is used to select the objects with the highest similarity in the p-th frame and the q-th frame according to the similarity and classify them into one category to form the trajectory of the same vehicle. 2. The expressway traffic event automatic detection system according to claim 1, wherein the vehicle state discrimination module specifically comprises: The speeding discrimination sub-module is used to judge the vehicle target whose speed exceeds the maximum speed limit threshold in the same vehicle target trajectory as speeding; The low-speed discrimination sub-module is used to determine a vehicle target whose speed is lower than the minimum speed limit threshold in the same vehicle target trajectory as a low speed; Violation of parking discrimination sub-module, used in the same vehicle target track, the existence of a vehicle target whose speed is zero is determined as a violation of parking; The retrograde discrimination sub-module is used to judge the vehicle target whose speed is negative in the same vehicle target trajectory as retrograde; Illegal occupancy of the emergency lane identification sub-module, used to determine the illegal occupancy of the emergency lane when the vehicle target angle is within the angle range of the emergency lane in the same vehicle target trajectory. The angle range of the emergency lane needs to be determined according to the installation angle of the millimeter wave radar equipment calibration; The frequent lane change discrimination sub-module is used to judge the frequent lane change when the vehicle target angle is continuously changed to different lane angle ranges within the detection range in the same vehicle target trajectory, and the lane angle range needs to be determined according to Millimeter wave radar equipment installation angle calibration. 3. The expressway traffic event automatic detection system according to claim 1, wherein the vehicle feature extraction module specifically includes: The image fusion sub-module is used to match the images collected by the millimeter-wave radar and the images collected by the video, and unify the coordinate system; The projection sub-module is used to detect the vehicle target information on the millimeter wave radar, project it in the image collected by the video, and identify the interest point area, and the interest point area represents the position of the vehicle target; The feature recognition sub-module is used to define the vehicle target extraction boundary according to the point of interest area, and identify the vehicle features within the extraction boundary, including license plate identity features, body features, vehicle type and body color; The correlation matching sub-module is used to correlate and match the vehicle feature information used for video image recognition with the trajectory information recognized by the millimeter-wave radar, store it in the processing unit, and upload it to the background management unit in real time. 4. The automatic detection system for highway traffic incidents according to claim 1, wherein the background management unit includes a data display platform, a system operation and maintenance management platform, and an abnormal event review management platform. 5. The expressway traffic event automatic detection system according to claim 4, characterized in that, the data display platform is used for combining big data visualization and GIS applications to report the system summary, monitoring point distribution, and real-time traffic events , traffic event statistics, traffic flow status, video surveillance status, and radar detection target track display status for dynamic display; the GIS means a geographic information system. 6. The expressway traffic event automatic detection system according to claim 4, characterized in that, the system operation and maintenance management platform is used for intersection radar and video detection equipment, processing unit equipment, broadcasting equipment, display equipment, network Online management of equipment, including equipment basic information maintenance, equipment failure alarm or personnel authority management. 7. The expressway traffic event automatic detection system according to claim 4, wherein the abnormal event review management platform is used to review and enforce traffic abnormal events collected on-site, according to the event type reported by the millimeter wave radar , as well as video evidence collection information collected by video equipment, conduct manual review, clarify illegal vehicles, types of violations, time and penalty standards, and transmit the obtained information to the traffic control platform for law enforcement.

Images