Automatic Searching of Optimal Bandwidth for Human Motion Tracking of Mean Shift Algorithm

The traditional Mean shift algorithm is a fixed bandwidth which leads to poor tracking performance even failure when the body is moving away from or gradually close to the camera movement.Taking into account the binary image after the body moving target feature extraction,the head area of the image presents a circular shape,and Hough transform can detect the size of the radius.In this paper,proposing an automatic bandw idth selection algorithm,which is according to the scaling of the radius based on the Hough transformation in the video sequences of the human body.The experimental results and analysis show that the algorithm has strong accuracy and robustness.