Study Of Target Detection And Tracking In Image Sequences

The modern war is going into new form war which is based information and communication. Orientation of target in war is more and more exact, attacking target for the real-time nature is increasingly high. So it is important for moving target to tracking and detect. This paper mainly researches detection and tracking methods of dim target. It is difficult for moving dim target to tracking and detect because the Signal Noise Ratio(SNR) of target is low and it has not texture etc.First of all, this paper introduces the state of researches of dim target detection and tracking in domestic and abroad. And, it describes some aspects which is increasing image, detecting target, tracking target, etc. Then, it introduces some elementary methods which is about increasing image, detecting target and tracking target.Increasing image is mainly introduced into two aspects which are space domain and time domain. As to detecting target, this paper just introduces mathematical morphology because the methods about detecting target are too much. The methods about tracking target are introduced into two aspects which are gate tracking and correlative tracking.Secondly, a new method of dim target detection and tracking based the feature of target area is proposed. This method takes out some features from target area, such as mean, variance. Then, it dilates the points which fit in with the features. After that, we can match the two images which are the maximum mean points'image and the minimum variance points'one so as to get the common area which is the target area. It can orient the target area quickly and exactly. Then, it can track the target more easily.Finally, the method of dim small target detection and tracking based on the wavelet transform and mathematical morphology is described, which increases the image of target and gets rid of the noise which is high frequency, then, making use of information of inter-frame, it tracks the target more effective. Then, we can correlate the inter-frame information so as to prove the target area is correct. After that, we use the tracking method which can predict the trace of target. It solutes the cases which are losing target or the target is sheltered. Simulation results prove the validity of these methods.