文档介绍：Abstract
Automatic face recognition is a cross-study problem puter vision, pattern recognition, artificial intelligence, psychology and other disciplines, the automatic identity authentication technology based on faces has broad application prospects in the field of security and entertainment. Therefore, doing automatic face recognition research has a scientific and practical significance. The automatic face recognition technologies based on the two-dimensional image have encountered problems such as illumination, pose and expression changes which are difficult to be e.
For the data acquisition mode, 3D face data acquisition mode is free from the impact of the face pose and changes in ambient light, face recognition algorithms Based on 3D faces data have the potential to e the difficult. Focused on three-dimensional face pose change and the issue of face recognition with varying expression, in this paper, an effective solution for the three-dimensional face pose change is presented, a new face representation methods is proposed for the issues of face recognition with varying expression. Related experiments are done and experiment results are analyzed and discussed. The main work is summarized as follows:
A thorough survey of the-state-of-the-art in 3D face recognition research is presented.
In this paper, some monly used concept in face recognition, as well as the typical processes included in the framework of face recognition system are introduced in details. And some relevant technologies are also discussed. Then, based on the forms of the face of the original data represented, 3D face recognition algorithms have been classified and the performances of some typical algorithms are listed.
Based on facial feature points and the geometric characteristics of the face, a three-dimensional face pose normalization method is proposed.
Using automatic detection of 3D facial five key points coordinates on the face , that is the two inner corners of two eyes, bridge of