【摘要】 通过提出一种基于奇异值向量和奇异值的人脸识别新方法——正交奇异值方法,将奇异向量 和奇异值中的信息有效结合,克服了目前基于奇异值分解的识别方法中,要么丢弃了奇异值中的信息, 要么丢弃了奇异向量中的信息,识别效果并不理想的缺陷。在ORL国际标准人脸库上实验显示,与目前 基于奇异值分解的识别方法相较,该方法达到了更好的识别效果。更多还原

【Abstract】 Methods of recognizing singular values or singular vector projection coefficients of SVD as face fea-tures are proposed and used popularly. However, these methods discard the information either in singular vectors or in singular values. In order to preserve this two kinds of information,we propose a new method which can preserve both of them.The ORL face database is used in experiment and experiment results indicate that the novel method is effective and superior to many existing methods.更多还原