Design And Implementation Of Regional Target Detection And Joint Tracking In Intelligent Video Surveillance System

With the rapid development of information technology in today’s society, intelligent video surveillance has been widely used in all fields of life. Regional target detection and joint tracking is an important research direction of intelligent video surveillance. It uses digital image processing technology for detecting and tracking moving object, and on this basis to match object. Using the system, monitors can deal with the emergency situation effectively. In addition, they can also save lots of manpower, material and financial resources, greatly reducing the rate of false positives and false negatives.The paper studies some key technologies of intel igent video surveillance about target detection, tracking and matching, and proposes some improvement comparing with the effect of the past methods. Furthermore, the research results are applied to pratical projects. At first, extracting the moving foreground object. Using the improved vibe algotithm to extract moving object, who runs very quickly and can restrain the impact of noise, and the improved vibe algotithm is able to remove the ‘ghost’ area quickly. Then the next work is to remove the shadow of moving target. Second, for the extracted moving object, the paper uses the area tracking method based on multi- feature matching and kalman filter, which can effectively track the target in the monitor ing scene. At last, the final step is to match the target. Because the effect of single feature for matching is often not ideal, so the paper uses an improved multi- feature fusion method. Through removing the interference of the background factors, combining color distribution features and local SIFT features to match target, of which the effect is pretty good.Finally, the paper has applied the above methods in a real project. According to a video surveillance scene, we analysis experimental results.