Feature extraction of three dimensional (3D) facial landmarks using Spin Image

摘要

3D anthropometric data can be obtained more easily than several years ago, but to process 3D data is most challenging and local feature extraction is one of the challenges. To extract local features of a landmark on 3D anthropometric data, numerous methods have been proposed. But the method with high efficiency, good robustness and strong adaptability is still not found out. In this paper, an approach using spin image was adopted to describe the 3D facial landmark features. Two hundred 3D head data of young male Chinese soldiers collected by a Chinese military institute in 2002 were analyzed. Preliminary results show that spin image is highly efficient and robust to pose and lighting. Further investigation is necessary to evaluate the feature descriptive ability of Spin Image in future 3D landmark automatic identification.