基于散度差的彩色人脸图像统计正交分析方法

由于彩色人脸图像比灰度人脸图像包含了更多的信息,彩色人脸图像识别方法越来越受到学者的重视。而对于研究最多的RGB彩色空间,通常R（红）、G（绿）、B（蓝）三分量间存在很大的相关性。为了最大程度的去除各个分量之间的相关性从而提高识别效果,有学者提出了基于统计正交投影变换（SOA）的彩色人脸图像识别方法。然而,该方法在特征提取的过程中不可避免的存在奇异性问题。为了解决这个问题,本文提出了一种基于散度差的彩色人脸图像统计正交分析方法ScatterDifference based Color Face Image Statistically Orthogonal Analysis Approach （SDFSOA）。此外,我们对所涉及的参数进行了合理的设置。实验表明所提方法能取得更好的识别效果。

Statistically Orthogonal Analysis Approach for Color Face Image Based on Scatter Difference

Color face image contains more information than gray- scale face image,thus color face recognition is attracting more and more researchers＇ attention. The most popular color space is RGB space. Usually,there exists much correlation between R（ Red）,G（ Green） and B（ Blue） components. In order to remove the correlation among features of three components for improving recognition results,a color face recognition method based on Statistically Orthogonal Analysis（ SOA） of projection transforms has been presented.However,this method suffers from problem of singularity inevitably. To solve the problem,propose a novel color face recognition approach that is Scatter Difference based color Face image Statistically Orthogonal Analysis（ SDFSOA） approach. In addition,set the parameter in SDFSOA reasonably. The experimental results indicate that the proposed approach can obtain favorable recognition results.