Moving Object Detection And Tracking In Video Surveillance System

With the development of computer technology, more and more video surveillance systems have been widely applied in production and life, such as the security of community and important site, the traffic security of highway, the military target tracking, intelligent weapons and so on, which play very important roles in many sides of society. Moving object detection and tracking is the most important part in video surveillance system. However, because of the complexity of the illumination condition and the object modeling of the real world, this problem becomes a present hot research.This paper proposes a detection method of moving object in varying indoor environment based on background Gaussian model. Firstly, the background is modeled, object can be detected according to the confidence interval of background model, feature parameter and brightness distortion of object, then the model and parameters of segmented object are renewed, and, are used as the detection basis of the next frame image combining with the gradient difference information of neighboring frame to resist the continuous varying of object and background along with time. Experiments show this method could overcome the varying of scene and track object effectively. As to outdoors environment, all pixels of picture are divided into background pixel, foreground pixel and mixed pixel based on inter-frame moving information. Background and target are modeled separately. According to the parameters of these models, the mixed pixels can be determined as background pixel or target pixel; finally the target can be detected. In the follow-up frames, the background model is utilized to detect targets, and then updated according to the detected result, so as to resist the change of the background and realize moving object detecting in the outdoor environment.