Complex Context Of Real-time Target Tracking And Recognition Research

Xu Rui-xin (Mechanic and Electronic Engineering)Directed by Liu Wei-ningReal-time tracking moving objects in complex background is a study field of Automatic Target Recognition (ATR), it include segment target, pick-up character and recognize target. In practical application, the operating time of tracking algorithm must be also considered In this paper we first introduce some correlative contents, essential algorithm and development of Automatic Target Recognition, then we discuss some algorithms of image segmentation and image matching. Different accelerating algorithms of image matching are studied. Finally, we proposed a real-time tracking algorithm based on cut model in order to overcome some shortcomings of correlation-matching algorithm. An Adaptive Model-Based algorithm is proposed, which can change model size while the focus of camera is changed. Process of matching is accelerated by pyramid algorithm in order to increase speed. Experiment results show that the improved correlation-matching algorithm has better matching precision and real time.In this paper, CCD grabbed the tracked images and all algorithms were accomplished by VC++ in Windows NT environment.