Moving Object Detection And Identification Technology Research Based On The Video

The detection and identification of the moving object is one of the hottest issues in computer vision area, and it is the basis of establishing some systems, such as intelligent monitor, intelligent transport system and so on. This paper emphasizes on studying the technology of detecting and identifying moving objects including detection of moving objects, the remove of shadow, the identification of moving objects and so on based on computer vision processing technology. The mainly finished research works are as follows:(1) Study and put forward to a way of detecting moving objects which combines the thought of background difference and interframe difference. To identify the final object, this method makes full use of last frame's detection result of moving objects, and combines with the result of current frame which is got from the thought of interframe difference. This method can detect the moving object partly in current image well, and adjust to suddenly change of environment; it can also avoid the problem of "double image" and "inanity" which may be occurred in the traditional interframe detection process.(2) Study and put forward to a method of shadow detection and elimination which is based on the thought of fuzzy cluster and edge detection. Firstly, the properties of shadow are studied; the method which combines fuzzy cluster and edge detection to eliminate the shadow of moving objects makes full use of a property of shadow that is the intensity ratio of moving object's shadow parts and the shadow parts covered by it always varies in a certain range. Experimental results show that, even if the intensity of moving objects and shadow are very similar, the shadow part can still be detected well.(3) Quickly identify the objectives is one of the most important part in motion detection of the target for the application. The method of based on the center of hausdorff is adopted to identify the target by matching the templates. The method resolves the problem of the sensitive to noises and timely not high of the traditional method. Moreover, the administration of the templates is also an important part of the matching. In this paper, the details of the database constructed and management are introduced.