The Study Of Nightly Surveillance System

The intelligent video surveillance system, a new kind of digital surveillance system based on the analysis and procession of video content, is playing an increasingly important role in the national economy. The key technology includes motion detection and background updating. One of primary complications deteriorating the stability of a night monitoring system is lighting variation. This paper studies on this problem. Background subtraction and frame difference method are commonly used moving objects detection methods. Background subtraction is simple but liable to be affected by lighting variation. Frame difference method is robust to lighting variation but difficult to segment and receive a whole moving object. Aimed at this instance, in this paper, a new motion detection method is developed by combining the background subtraction and frame difference methods together with a background self-updating method, in which several image frames are processed comprehensively. This method bears not only the characteristic "exact to survey object" of background subtraction but also the characteristic "robust to lighting variations" of frame difference. Further considering the background self-updating arithmetic, the proposed method is supposed to be able to improve obviously the stability of a video surveillance system, especially one used at night. Some pre treatments, like Convolute-Simple filtering and morphologic operation etc., have also been performed to the images after subtraction to solve noise-disturbing problem, further enhancing the stability and security of a video surveillance system. To validate the feasibility of the proposed detection and tracking method, a prototype system has been built and tested in different environments. Experiment results show that the presented method is effective to conquer the infection of lighting variation. The system could resume stable status in about 0.6s after lighting varies intensively, so it has the very high practical value in visual surveillance.