A Study On Single Object Tracking Method

Video surveillance just use image process and computer vision detection technology, through for surveillance video recording automatically and timely analyzing, get target location, recognition and tracking in dynamic scene, and analysis and judge target behavior on this foundation. The orders of bank window depend on persons who carry out business and securities. The tracking system of actor on the stage depend cameramen artificially. They also don't reach intelligent. The paper mainly aims at the two situations and studies target detecting and tracking algorithm.Target detecting, first, through frames difference based adaptive background update algorithm construct background model and real-time update; After for the current frame image adopt the background image difference method to calculate, the differential image carry through gray morphological filter and carry on image segmentation, finally, carry through shadow examination in the hsv, and remove it in segmenting image. The experiment result certificates, this algorithm solves the problem that smally colorific difference between suit color of people and color of ground as well as the effects generated by natural light, additional light and shadow to the recognition, it can recognize moving target accurately.Target tracking, First, through constructing background model and background image difference method detect motion target; After for the point of the target construct gray probability density function; similarity measurement is determined by the distance defined by Bhattacharyya coefficient; finally with meanshift carrying on matching area search; or a prediction based on a Kalman filter is adopted to target tracking. The experiment results show that with the algorithms, the tracking basically satisfy timely and accurately.The algorithm easily software realize, compute quantity little, practical effective well. For vedio surveillance in the fixed scene and target tracking in dynamic scene, it have definite applied value. The experiment prove that it may apply monitoring and detecting alarm system for people crossing the line in queue to the bank window and danced actor automatic tracking on the arena.