一种有效的人脸识别方法

AN EFFICIENT RECOGNITION APPROACH TO HUMAN FACES

人脸识别是模式识别领域中一个相当困难而又有重要理论价值的研究课题。本文首先引入表示人脸的特征图像和投影图像的概念,提出利用投影图像在特征图像上投影的坐标作为描述人像本质属性的特征矢量,并导出了基于投影图像的人脸特征抽取方法。最后构造了一个层次的距离分类器进行人脸的识别。实验结果表明基于投影图像的特征矢量具有很好的稳定性、鉴别能力和识别率。

Human face recognition is a very difficult and important problem in the field of pattern recognition. In this paper, the characteristic images and projective images are first introduced. The projective coordinates of projective image on the characteristic images are presented to describe the intrinsic attributes of human faces, Then, a feature extraction method of human faces based on the projective images is proposed. Finally, a hierarchical distance classifier is designed to recognize human faces. Experimental results show that the feature vector based on the projective images has good stability, discriminant ability and recognition rate.