Face Contour Extraction and Tracking Using Level Sets
Huang Fu · Chinese Journal of Computers · 2003
In this paper, a novel method using level set technique for face contour extraction and tracking in image sequences with simple backgrounds is proposed. Firstly the motion boundary is quickly detected via interframe image difference. After appropriate thresholding and morphological operations, the possible face region is then found out according to heuristic rules in face profile projections. Using the detected possible face region rectangle as the initial evolving curve, the accurate face contour is extracted using a modified level set model. Assuming that the face is always moving in image sequences, a first order Kalman filter is then used for motion estimation. Thus the face contour can be tracked with the same level set model in the following image frames. The experimental results show the efficiency of our method.