A Novel Vehicle Tracking Algorithm Using Video Image Processing

Due to convenient installation, large coverage area and be effective, video detection technology has become a research focus in current traffic incident detection field. This paper simply introduces common tracking algorithm and its principle of several moving objects and puts forward a Kalman filter-based feature matching tracking method. First, we apply temporal difference method to detect moving objects and obtain their initial position. Then, this paper adopts Kalman filter to predict the moving objects position at next period and utilizes results of moving objects detection to evaluate and correct predicting results. This obtains correct position of moving objects and analyzes vehicles movement according to tracking results. The experiment results show this method can effectively solve reliable tracking under condition of partial occlusion and short-time total occlusion of moving targets.