基于局部特征的三维人脸识别

3D Face Recognition Based on Local Feature

Shape Index和曲度用于构造局部特征,并运用于三维人脸识别中。这种局部特征的提取方法用于人脸识别,不需要预先进行对齐处理,而且对有遮挡的人脸数据具有相对较好的识别效果。这里,局部特征提取的主要步骤如下:在不同的尺度上对三维人脸上的关键点进行检测;再对检测到的关键点确定主方向,然后根据主方向构造关键点在某一邻域内的特征向量。实验所用数据库是Bosphorus Database,是Bogazici大学采集的三维人脸数据库。

Shape Index and curvature is used to construct the local characteristics, and applied to 3D face recognition. This kind of local feature extraction method for face recognition, don＇t need to align in advance, and also have relatively good recognition rate under external occlusion. Here, the main steps of local feature extraction are as follows： Firstly, test the key points of the three dimensional person face in different scales; After that, determine the reference direction of the key point, then construct a feature vector of key point within a neighbor according to the reference direction. Experiment data is come from Bosphorus Database collected by the researchers of the University of Bogazici.