Research And Application Of Intelligent Video Surveillance System Based On Fixed Scene

With the rapid development of computer image and visual treatment technology, intelligent video surveillance gradually become one of the closely-watched foremost subjects. Intelligent video surveillance is a technology that in the situation of needing no intervene of peephole, utilizing computer vision and video image analytical technique to analyze image sequence by itself, to detect and monitor the target in scene automatically, and then to analyze and judge the target's behavior. This paper analyzes the existing algorithms based on the industry, making the video surveillance more on-line, accurately, through researching and improving the intelligent monitoring system of target detection and target tracking algorithm. Finally, improved the algorithm of the entire system.The target detection is the basis for the intelligent video surveillance system. By analyzing the background difference method and the advantages and disadvantages of frame difference method, this paper chose a limited frame means algorithm and background subtraction method to obtain binary image according to the characteristics of a fixed scene. Then, using a new algorithm to mark the region based on regional growth combined with connectivity of connected components labeling. The algorithm for the segment markers and the advantages of regional growth can be quickly and accurately separated and labeled on the scene more targets, and has greatly improved in the time complexity than the previous algorithm.Tracking of the moving object is a hot topic for monitoring system, but also is an important research in the paper. By analyzing variety of tracking algorithms presented in the profession, this paper is focusing on the Kalman and Mean-shift filtering algorithm. Kalman estimate the location of the image in the next frame of the target filtering from the macro level; while Mean-shift algorithm searches for the target in the optimal position in local area. This paper combines the above two algorithms to improve the target tracking algorithm in order to solve the problem that the target occluding and appearing a short time away. The results show that the improved algorithm is better.