Research On Moving Object Detection And Shadow Suppression

With the development of computer technology and image processing technology in recent decades, target detection system with moving object video is served to many fields of defense as well as national defense and civil economic construction. With application fields expanding, people have made more and more in-depth researches about moving target detection and shadow suppression. People proposed a variety of effective methods. The target detection and shadow suppression of moving object have become one of the hot research topics for application prospects and potential economic value in the field of intelligence surveillance.First of all, the thesis summarizes the research background and significance of the topic, and briefly introduces some classical theory and method of moving target detection, then analyzes and compares the advantage and disadvantage of the proposed detection methods. It pays attention to the Inter-frame difference method, optical flow method, background difference method as well as Gaussian model method. Then it discusses the basic theory and realization of above methods, some experimental data are also proposed.Secondly, the thesis mainly analyses Gaussian model method. As it is not accurate to acquire the first frame and it is too slow to acquire stable frame using traditional Gaussian model method, a new method combining of median method and majority method is proposed to improve the accuracy of first frame. The thesis use a new parameter reflecting the match of current frame and Gaussian background model to improve the updating of background model and to avoid the slow object melting into background.Besides, according to large consumption in time and space using traditional Gaussian model method, the thesis propose an adaptive selection strategy of number of components of mixture of Gaussian model to improve the efficiency of modeling. The new method reduce the consumption in time and space, as well as build the background model fast for moving object detection.Finally, the existence of shadow causes interference to moving object detection because of the effect of illumination. The thesis proposes a new shadow suppression method with no threshold to detect moving object with shadow. Compared with the traditional shadows suppression method, this method can suppress shadows to moving objects without setting the threshold value.