Vehicle monitoring technology based on the video image due to its features of easy maintenance and easy implement is concerned in recent years. Vehicle tracking is one of the key technologies to achieve video vehicle monitoring. How to achieve real-time vehicle tracking is the key and difficult problem at home and abroad. A method of Kalman tracking based on virtual loop is proposed in the paper. Two parts about vehicle presence detection and vehicle forecast tracking are introduced in this paper. Experimental results show that this method in the fast-path scenarios can effectively achieve vehicles tracking, fully meet the system requirements and have high practicality.
Since 1980s, with the development of scientific and technological and the social progress, the urban population increased, the number of cars is increasing, The road has been unable to meet growing traffic’s demand, so traffic congestion, traffic jams and accidents are increasing. To solve these problems, we need a powerful means of traffic management to control traffic and avoid traffic jams, but management systems rely on human is low efficiency and high cost. That apparently no longer remission the traffic situation. With the rapid development of multimedia technology, video has become the main carrier of information dissemination. Therefore, vehicle monitoring system based on video image is the trends of the world’s road traffic. In order to achieve video traffic monitoring, real-time vehicle tracking and stability are two key factors. In this paper, the virtual detector based on tracking by estimating the vehicle’s position is proposed, this method only needs to extract the background image, then set the virtual loops to detect vehicle’s state. There is a method established to tracking the vehicle by estimating the states of the virtual loops. Experimental results show that the method can improve the stability and real-time of vehicle tracking.
n the traffic monitoring system, the aim of vehicle tracking is to determine vehicle trajectory, then we get corresponding relation between foreground object and dynamic object. In video sequences, the time interval between adjacent two images is small, so the mutations of vehicles motion states are not happen. Therefore, it can be considered that the tracking window’s size and position of the target vehicle have little change. Code Shoppy So the target vehicle centroid and tracking window’s size can be used as the characteristic to track, then using the Kalman filtering forecast the target vehicle state in next time, it can be minimize the target area. We get the measure parameters while matching the characteristic values. Then update motion model using Kalman filter, and finally form a tracking chain to get vehicle trajectory.
This paper presents a predictive tracking method based on the virtual detector, and introduces their theoretical derivation and implementation process in detail The tracking method was tested ,and the test data show that the method is valid in the vehicle detection system ,so this method has great practical value.