Software Design And Implementation Of Abnormal Behavior Detection In Video Surveillance System

At present, with the development of economy and the improvement of people’s safety awareness, the video surveillance system in various occasions, such as: Street, square, residential and so on. Video surveillance system is the significant effect on the protection of public safety. Most of the current video surveillance system has been unable to meet the needs of social development, intelligent video surveillance system will become the main trend of development in the future. More and more researchers begin to research and development of intelligent video surveillance system, and on the human action recognition in video surveillance has become a hot research topic in recent years.At present, most research is still in the simple act of detection to the human body, the research of complex movement is less, but in the video surveillance system, of complex movements was the more need.The paper mainly researched the human abnormal behavior detected, focus on complex behaviors such as robbery, fighting for the detection of abnormal behavior. First of all, introduced the current situation in the domestic and foreign development of intelligent surveillance system, then introduced the algorithm of computer vision and video monitoring system with some achievements. In this paper, on the basis of the proposed method for moving target detection using background subtraction, and increased the adaptive function of the background and application of image processing technique for denoising. The prospect of the target image and then extracted for moving target tracking by pyramid Lucas-kanade(LK) optical flow method. This paper putted forward to used direction- amplitude histogram to described of human behavior. In the histogram, maxima interval velocity was calculated to detect fast behavior. Introducing the concept of entropy and variance was used to detect the fighting behavior through computing direction entropy, amplitude entropy, velocity variance. And through the CASIA database on the algorithm in this paper test, used the video from the perspective of different shooting to verify the effectiveness of this proposed algorithm, and the high accuracy.The algorithm first were validated by OpenCV( Open Source Computer Vision Library), and then combined MFC( Microsoft Foundation Classes) with OpenCV to design a human abnormal behavior detected software, and test the result of the software, realized the real-time analysis and processing of the video data, and the subsequent function development of the monitoring system can also be based on this software.