Research Of Motion Detection Algorithm Base On Outdoor Video Surveillance

According to the motion detection problems in intelligent video surveillance system, this paper makes a thorough study into pre-processing of the target detection algorithm, optical flow and VIBE (Visual Background Extractor) method of moving target detection technology. Though the analysis of the image by image scaling, image smoothing and image gray processing, and then using a variety of smoothing algorithms for the process of the image, the experimental results showed that the median filter can remove salt noise and Gaussian noise effectively. The proposed method, Inter-Frame Regional Corners Difference algorithm, can eliminate static state background and reduce the amount of optical flow calculation effectively. We analyze the vector of the flow and apply the vector movement restraint, the results show that the proposed method can reduce dynamic background disturbance and extract the foreground effectively. On study of the VIBE algorithm, then the author proposed an improved method on the basis of VIBE algorithm in the different of outdoor surveillance scene. The experimental tests show that the improved algorithm is more robust, flexible, and adaptive in outdoor scenes. The shadows of foreground objects in video surveillance often result in the following problems: fusion of the moving target, geometrical characteristics change, false targets and other issues. And research shows that it cannot remove the shadows accurately when the proposed method just relies on a kind of shadow removal algorithm, in this paper, a method of combing the characteristics of the pixel brightness, pixel texture characteristics and color ratio information is proposed, which can obtain the better results. The experiments show that the algorithm can remove the shadow pixels and reserve the moving target prospects.