Tracking by detection of a human being in image sequences

Tracking a moving person in an image sequence is an important task for many video surveillance applications. In this paper, a tracking by detection method based on a background subtraction method and the Kalman Filter predictor is proposed. Firstly, we extract the human target silhouette automatically using a background subtraction method. Secondly, we predict the location of the extracted human silhouette in the next frame using the predictor Kalman Filter. After that, we obtain the observation in the current frame using the background subtraction method. Finally, we update the tracker system by the correction step of Kalman Filter. Numerous experiments on several challenging image sequences illustrate the effectiveness of the proposed method.