Real Time Face and Hand Tracking With Correlation

In this paper, we demonstrate an effective tracking system for moving face and hand objects in real time. Our approach contains of three substages: adaptive background subtraction, face and hand detection using skin color and tracking them with pixels correlations respectively. There are famous robust tracking methods such as mean shift and active shape model in the literature. Despite providing robust object tracking, these methods have high computational burden. Taking into account the requirements of online processing, our approach can detect face and hands in image sequence using their skin color, geometrical shape information, and prior probabilities. The effectiveness of our approach comes from its simplicity. We demonstrate the performance of the proposed method by comparing it with the traditional tracking methods on real sequences.