基于鉴别分析的彩色人脸图像灰度转换方法  

Color Facial Discriminate Feature Extraction and Recognition

作　　者：儒林[1] 

机构地区：[1]临沂职业学院,山东临沂276017

出　　处：《科技通报》2012年第4期94-96,共3页Bulletin of Science and Technology

摘　　要：当前主流的人脸识别算法,都是把原有的彩色图像转化为灰度图后,采用基于灰度图像的特征抽取与识别算法进行分类识别。人们在实际操作过程中,只是使用一组简单的加权系数实现从彩色图像到灰度图的转换,这并不能很好的体现R,G,B 3个颜色分量之间的次重关系。本文根据人脸图像颜色组成的特点,对彩色人脸图像的R,G,B 3个分量的颜色信息进行特征抽取与分析,从中找出鉴别特征的三基色系数表示方法,把彩色图像转化为灰度图。最后,在国际通用的AR标准彩色人脸库中进行了大量实验,验证了本文算法的有效性。Current popular algorithms for face recognition all share the same characteristic, they all convert the origin color image to gray image. After that, operate the class discrimination based on the feature extraction and recognition algorithm of the gray image. During practical operations, people convert color images to gray images with a simple combiration color coefficients. This can＇t tell which is more important of the three components RGB. According to the color components of the facial images, this paper seeks to find a optimal combination coefficient which contains the most color information of the image, by feature extraction and analysis of the R, G and B components of color facial images. Then operate PCA on the combined components. Finally, extensive experiments performed on the international and universal AR standard color face database verify the effectiveness of the proposed method.

关 键 词：彩色人脸识别 鉴别特征 主分量分析 特征抽取 

分 类 号：TP317[自动化与计算机技术—计算机软件与理论]