움직임, 영역, 형상정보의 혼성 매칭에 의한 사람 추적

In this paper, the part of tracking of the human which is appling the tracing and detecting of moving target using motion vector and difference image with weighted temporal. And the goal of this paper is making the whole algorithm of people tracking. People tracker is consist of four part which is in close connection with each other. The modules are motion detector, region tracker, human feature detector and shape tracker. Which is combination and exchanging of two different module is the result of pass the limit and the reliability is the product. Also, combination of the several algorithm is efficient means which is overcome the limit of module and is made low complexity. Human detector is show to keep strongly tracing when occlusion and low quality of video image goes up. The problem point of people tracker is firstly, in case of sudden light variation as the sun block by the cloud secondly, occlusion happen in detecting group of people. The next goal of this paper is developing a software for people tracker and structure which is possible for maintenance. It must be expand easily, using easily and apply to another tracking in project also be make possible tracking and recognizing correctly and searching a motion using the Indexing technology which has a whole information of people motion.