人脸图像光照预处理的一种新方法  

A new method of lighting preprocessing for facial image

作　　者：唐良宝[1] 李丽平[1] 

机构地区：[1]桂林电子科技大学机电工程学院,广西桂林541004

出　　处：《光学技术》2008年第S1期149-150,153,共3页Optical Technique

基　　金：广西区研究生教育创新计划资助项目(2006105950802M09)

摘　　要：人脸识别是图像分析和理解领域中最成功的应用之一,近年来得到了迅速的发展,但是阻碍人脸识别技术应用到实际中的瓶颈之一——光照问题,一直没能得到很好的解决。局部二值模式是最近发展起来的一种理论简单但功能强大的纹理分析算法,在计算机视觉等领域表现出良好的性能。将该纹理提取算法应用到图像预处理中并并利用大规模中国人脸图像数据库CAS-PEAL-R1来检验这种方法的有效性。实验结果表明:加入LBP纹理后,该方法能较好解决光照变化问题,提高识别性能。Face recognition is one of the most successful applications in image analysis and understanding.It has received rapid development recently.However illumination variation is still one bottleneck preventing face recognition technique from being used into practice.Local Binary Patterns(LBP) is a texture analysis algorithms which is simple but its function is powerful.This method is developed recently.It has favorable performance in computer vision.The texture extract algorithms has been applied to the image preprocessing and the efficiency of the proposed approach is tested This paper presents Local Binary Patterns(LBP) as an facial image preprocessing to solve the variation of illumination.The efficiency of the proposed approach is empirically demonstrated which based on the database: CAS-PEAL-R1.The results show that our method can better solve the problem of lighting variation,and improve the recognition performance.

关 键 词：人脸识别 局部二值模式 光照变化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]