Research On Target Tracking Of Moving Background

The technology of video based object detection and tracking is one of the hotspots in the field of computer vision, which is also the basic and important technology in the application of smart surveillance, human-machine interface, mobile Robert navigation, full-sight caterpillar system and so on. This thesis focuses on the solution to the problems of the objects, especially motorcars, the detection and tracking in the application of smart surveillance, including target detection and tracking algorithm, tracking algorithm's acceleration and registration algorithm. Main achievements of this dissertation are as follows:(1) An approach to detect and track multi-target in a static background sequence was proposed. It was based on local recursive Otsu segmentation and close operation. The algorithm settled the single-target-divided problem effectively, and prevented targets'conglutination by choosing right close template.(2) Faced with the problem that it was liable to appear pseudo-target in low contrast moving background conditions. The paper proposed the least two-dimensional fitting of cross-correlation algorithm to pick out the pseudo-target.(3) A particular match-template-refreshing method was applied in order to settle the problem of dropping out the target cased by the target's size-changing.(4) A weight-counting plate was presented to solve the problem that when the target was sheltered by other object the system always lost the target. The new algorithm gave the middle of the target a high weighting, and not refresh the template as frequently as the original match-template-refreshing method. The new method settled the shelter problem in a moving background followed by the size changing of target effectively.