摘要
提出了一种人脸识别图像的光照纠正预处理算法。该算法首先将均衡化引入仿射变换算法和光照补偿算法中,改进它们的处理效果;然后采用一种基于直方图灰度重心分布统计的方法,将多种预处理算法有效地融合在一起,得到新的光照纠正预处理算法。实验表明该算法具有光照无关性,能取得良好的光照纠正效果,并可大幅度提高复杂光照环境下的人脸识别率。
A new preprocess algorithm was presented to correct the illumination of face images. In the algorithm, histogram equalization was introduced into AT algorithm and ICR algorithm to improve the performances of them. And method based on the statistic of distribution information of histograms＇ gray barycenter, was also adopted to combine several preproeess algorithms effectively. Experiment shows that the new algorithm is illumination invariant, which can achieve better preprocess result and can increase recognition rate of face images greatly in complex illumination environment.
出处
《计算机工程与设计》
CSCD
北大核心
2006年第9期1547-1549,1559,共4页
Computer Engineering and Design
基金
广州市科技计划基金项目(2002XGP16)
广州市天河区科技计划基金项目(2002XGP11)
关键词
人脸识别
预处理
仿射变换算法
光照补偿算法
光照无关
face recognition
preprocess
affine transform algorithm
illumination compensation based on multiple regression model algorithm
illumination invariant
