Eye Feature Extraction Based on 3D Deformable Template

Wanjun Huang     Baocai Yin     Tongbo Chen     Dehui Kong    

Abstract

Eye feature extraction has become one of important issues in the applications of facial image perception. A new approach for eye feature extraction based on 3D deformable template has been proposed to extract the exact eye features of the facial images with large vertical rotation angle. The approach includes two new techniques. One is pose estimation strategy that can be used to obtain the vertical rotation angle of the facial image, and the other one is 3D deformable template matching which can be employed to extract the eye features of the facial image. The experimental results demonstrate the efficiency of the approach for both frontal facial image and the facial image with large vertical rotation angle.

Paper

Wanjun Huang, Baocai Yin, Tongbo Chen, and Dehui Kong. Eye feature extraction based on 3D deformable template. Journal of Computer Research and Development, 39(4):495-501, 2002. PDF

---