Research On Retrograde Traffic Incident Detection Method Based On Video Image

Traffic Incident Detection can discover a traffic accident timely,which can effective dredge the traffic to achieve the aim of traffic incidents warning and the reduce of traffic accidents. Traffic incident detection monitoring system will be able to achieve the aim of efficient and fast road traffic information monitoring, which will make the traffic control and the traffic processing more efficient.The paper deeply researched the retrograde traffic incident detection algorithm and moving target tracking, which was based on the current achievements. Retrograde traffic incident detection method is improved,which is based on object detection and object tracking.A retrograde traffic incident detection system is designed to verify the effectiveness of the proposed algorithm. Specific methods are as follows:1.By analysis of the different characteristic from frame difference method, optical flow method, background subtraction method, drew the advantages and disadvantages of these types of detection methods.In connection with traffic incident shadow which will be exist in the target,the paper improved the traditional Gaussian mixture model.2.By analyzing the existing algorithm, proposed a algorithm based on Kalman algorithm and Mean-shift algorithm.3.In the detection of the retrograde event, the mathematical modeling of the target tracking trajectory is firstly carried out, and the trajectory analysis is realized according to the angle between the straight line fitting lines. The relations between the moving direction and the direction of the moving target are judged according to the difference of the center point of the target. In order to prevent false positives, you can set a threshold value.Event occurred retrograde when a final determination is greater than a threshold. Experiments show that this method can accurately detect retrograde behavior of pedestrians and vehicles.