The Behavior Analysis Of Moving Objects In Public Scene

The behavior analysis of moving objects is an advance stage in intelligence surveillance. In this paper, we main attention to parameter the scene knowledge, detect the moving objects, track the moving objects and analyze its behavior. In order to faster and more accurately detect the linear boundary in public scene, we have compared Hough transform and Radon transform, and Radon transform was used to detect the linear boundary. For fitting the curved boundary, the cubic polynomial was cited. To detect the moving objects, the adaptive background updating algorithm and maximum variance adaptive thresholding algorithm were cited. In order to classify and track the moving objects effectively, the Hu moment feature of moving objects was used. In the stage of moving objects’behavior analysis and understanding, it was developed from two aspects, one is based on the boundary; and the second is based on the trajectory. Thus establishing the abnormal behavior detection model of moving objects in road scenes, and it realizes the intelligent control of traffic. The experimental results show that our algorithm is effective.