基于三维人脸成像系统的复数域人脸识别方法

Face recognition in complex domain based on 3D facial imaging system

三维人脸识别是模式识别和人工智能领域的研究热点。提出了一种利用人脸图像的二维灰度信息和三维深度信息进行人脸识别的算法。首先利用相关型图像传感器构成三维实时人脸成像系统,并将获得的三维人脸物理数据用复数形式表达。再将特征脸(Eigenface)方法和Fisherface方法拓展到复数域,提出了复数域特征脸方法和复数域Fisherface方法。改进的复数域人脸识别方法在三维实时成像系统的人脸数据库中进行的人脸识别实验表明,复数域的三维人脸识别方法明显优于传统的人脸识别方法。

3D face recognition is a frontier in pattern recognition and artificial intelligence area.A recognition al-gorithm with 2D intensity data and 3D depth map of face is presented in the paper.A 3D face real-time imaging system using CIS image sensor is built,and the 3D facial data（depth map combined with 2D intensity image）in complex do-main is described.Traditional Eigenface and Fisherface algorithms are extended into the complex domain,and Com-plex-Eigenface and Complex-Fisherface is presented.The experimental results show that the improved face recognition method in complex domain is better than the traditional face recognition methods.