Research On The Detection Method Of Continuous Motion State Of Vehicle Based On Video

In the traffic surveillance video, the vehicle motion state is completely recorded, which provides a prerequisite for video speed measurement. The continuous motion state of the vehicle can be used in traffic flow condition monitoring, driving behavior observation, traffic accident identification and so on. It has important significance to use the video to detect the continuous motion state of the vehicle.First this paper analyzes the research status of domestic and foreign vehicle speed measurement by video, in terms of obtaining the target location of different of calculation model were summarized. The advantages and disadvantages of different methods were summed up. Base on this, the research methods of extracting feature points from the vehicle itself was identified, and the goal of vehicle positioning was achieved through the tracking of feature points.Then through the analysis of traffic video vehicle running state, the feature points were extracted from the chosen characteristic region. The image recognition algorithm based on the weighted color histogram of wheel pixels is proposed, and the Mean Shift algorithm is used to realize the tracking of vehicle wheel heart. From the camera model, the relationship between the unit pixel of the front and back wheels in the image and the actual distance is analyzed, and the improved time interpolation method is proposed in this paper, based on this, the vehicle continuous velocity calculation model was established which is moving in a straight line.Based on the Ackerman triangle principle, the turning characteristics of the vehicle are studied. Based on this, the trajectory of the front and rear wheels is analyzed, and the formula of inner wheel differential expression is given. From the inner wheel, the performance of the inner wheel in the image is studied, and the calculation model of the continuous speed of the turning vehicle is established by using the time interpolation method.Then, the accuracy of the model was verified by real vehicle speed tests. The emergency braking speed test and the small radius curve speed test were designed, which verified the accuracy of the model.At last, the measurement system of the continuous velocity of the vehicle is developed. Based on the MATLAB platform the GUI human-computer interaction interface was designed, establishing the video playback and decomposition, feature selection and tracking, vehicle trajectory and continuous speed output functional modules, which realized the vehicle speed automatic measurement by using video.