Facial skin motion localization in a video sequence

We present in this paper facial skin motion localization in a video sequence. When a face is under motion, a set of muscles, lip shape, eyebrows... are involved. The changes in successive images have direct effects on facial motion. Our work proposes an approach to detect points and areas that undergoes significant movement on a face under motion. Our method requires only simple devices: a webcam and a PC, and our system needs minimal users interaction without the need for hardware and software modification or special placement. From a video scene, the first step of this work is to follow automatically the motion part of the head using the sigma-delta method. We use the tracking results to detecting the motion of the facial skin. The next step is to apply the method of Horn & Schunk to calculate the field velocity and orientation of points on the facial skin in order to compute significant displacement using the maximum likelihood.