摘要
提出了一种粗糙集与mPCA相结合的人脸识别算法。根据一定规则将人脸图像模块化,对每一个小模块利用PCA进行处理,对于经过PCA降维后的数据再利用粗糙集约减,去除冗余信息。该方法可以减少姿势表情的变化给人脸识别带来的影响,去除大量的冗余信息,从而降低计算的复杂性,提高识别率。基于ORL人脸数据库的实验结果表明,该算法正确识别率达到97%。
This paper proposes a new face recognition approach based on rough set and mPCA. It modularizes the face images according to some very rules; and deals with the modular images by using PCA, reduces the data which has been descented by PCA. This method not only reduces the influence to face recognition brought by the change of pose and expression of the face images, but also wipes off a lot of redundant data. Thereby, it reduces the complexity of computation and enhances the recognition rate observably. Experimental result based on ORL face database shows that the correct recognition ratio can reach 97 percent.
出处
《计算机工程》
CAS
CSCD
北大核心
2008年第14期179-181,共3页
Computer Engineering
关键词
粗糙集
属性约减
人脸识别
rough set
attributes reduction
face recognition