基于曲率和纹理信息的三维人脸特征提取

Facial feature extraction based on curvature and texture

针对目前三维人脸特征提取方法对表情很敏感的缺陷,通过借鉴曲面曲率生动描述人脸形状的特点和人脸特征区域纹理不受表情影响的特点,提出了一种基于曲率和纹理信息的三维人脸特征提取方法。实验分析说明该方法有效利用了三维人脸的深度信息和人脸纹理对表情和姿态不敏感,具有较好的准确性和鲁棒性。

To compensate for the limitation of the 3D facial feature extraction that is sensitive to the expressions, a new method of 3D facial extraction based on curvature and texture is proposed. This method uses the characteristics that the surface curvature describes the facial shape and the texture will not be affected by expressions. The experimental results show that this method which uses the range image of 3D face is not sensitive to expression and pose, and has high veracity and good robustness.