基于Top-hat变换与主成分分析的人脸识别方法

Face recognition method based on top-hat transform and PCA

针对光照条件对人脸识别性能的影响,提出了一种利用Top-hat变换对特征提取前的人脸图像进行预处理的方法,减弱了外界光照对人脸特征提取的影响,并将该方法与基于主成分分析的特征提取方法和基于多类别分类器支持向量机的分类策略相结合,形成完整的人脸辨识算法。最后在ORL人脸库和YALE_B人脸库上,对该方法的识别率进行了测试,并与传统的基于主成分分析和最小近邻分类方法结合的人脸辨识方法进行了比较,得到了较好的结果。

Aimed at the effect of the illumination condition to the performance of face recognition, a new method using top-hat transform is proposed to preproccess the image before the face feature is extracted, and the effection of external illumination to the extraction of face feature is deceased. The proposed method was combined with the PCA and SVM, and the integral face recognition method is obtained. Finally, on ORL face database and Yale_B face database, the distinguish ratio of the proposed method is tested, and is compared with the traditional face recognition method based on the PCA and the least near neighbor method, the better result is obtained.