3D ear modeling based on SFS

Ear recognition, by using ear for identification recognition, is a kind of new biometric identification technology. Currently there are a lot of ear recognition methods based on 2D images, while 3D data can provide more information. 3D reconstruction methods based on multiple 2D images have a common difficult which is to extract corresponding feature points of different images. In this paper, SFS(Shape From Shading) was used for 3D modeling by only one grayscale image. Light direction was estimated by analyzing singular points, which refers to the greatest grayscale point, in 2D grayscale image. In addition, in order to achieve more accurate 3D modes, the abnormal high brightness of the cavity of auricular concha is processed. The matching accuracy of the model can reach to 84%. Therefore the experiments shown that the method proposed in this paper is simple and effective to modeling 3D ear.