Moving Objects Detection Methods Research Based On Video Surveillance

Video surveillance is a current research topic which attracted more and more attention in image processing and computer vision. The higher level applications such as image segmentation, object tracking and behavior understanding depend heavily on the results of moving objects detection in video image. Effective detection of moving objects is a fundamental task in video surveillance and it is also a subject studied by a lot of scholars both at home and abroad. Background difference is the most common and effective method of detection and its key point is background modeling. This paper has done a lot of research based on the background modeling methods. The main task of this paper is as follows:1. An implementation algorithm of inter-frame difference is given based on the deeply research of the existing background model algorithms. Two consecutive frames are subtracted to pick up background pixels in this method, so, background frame is established. According to the background pixels obeying Gaussian distribution in time sequences, we update background frame. The experiments show that the algorithm performs well for attaining clean background without the influence of foreground objects and updates well for real time.2. According to the study of background model on the complex scene, a motion detection algorithm based on codebook is given. This method takes fully advantage of pixel's color and brightness information to build a codebook background for every pixel, and it has a good performance on moving objects detection. In order to remove the impact of shadow and improve the quality of detection, we use a shadow detection method based on vertical projection.3. Using open source vision library OpenCV in the Windows environment, this paper achieve a small moving objects detection system to perform and analyze the detection algorithm, and verify the effectiveness of the algorithm.