Face Tracking Using Unscented Kalman Filter

In this paper, we present an efficient vision-based face detection and tracking system. For face detection, we have improved the Viola-Jones algorithm by increasing the number of sample images for training. This improved cascade classifier is performing better than the standard algorithm. In this proposed method, we used the practical implementation of a face tracking system based on a nonlinear filter such as Unscented Kalman Filter (UKF). The experimental results indicate that real-time nonlinear tracking problem can be resolved by using this nonlinear Unscented Kalman Filter. This method is efficient in detecting moving faces and tracks them from a video without any apriori information about the captured scene.