基于Gabor特征量和核函数判决分析方法的人脸识别  

Face recognition based on Gabor features and kernel discriminant analysis method

作　　者：刘靖[1] 周激流[1] 

机构地区：[1]四川大学电子信息学院,四川成都610064

出　　处：《计算机应用》2005年第9期2131-2133,共3页journal of Computer Applications

摘　　要：研究了基于Gabor特征量和核函数判决方法的人脸识别方法,即首先利用Gabor滤波器组对输入样本进行处理,获得Gabor特征量;然后利用核函数判决方法实现人脸识别。Gabor滤波器组通过提取具有空间频率、空间位置和取向选择性的特征,较好克服了实际中由于表情和光照不同带来的变化;而核函数判决分析方法具有提取输入样本空间的非线性最佳鉴别特征的优点。实验仿真表明了该方法的有效性。A Gabor features based classification using kernel discriminant analysis for face recognition was introduced. First an augmented Gabor feature vector was derived from the Gabor filters. Then the kemel discriminant analysis （KDA） method was used for enhanced face recognition performance. The Gabor features characterized by spatial frequency, spatial locality, and orientation selectivity can do with the variations due to changes in illumination and facial expression. The KDA method can derive the nonlinear discriminant features in the input space. The feasibility of this method has been tested by experiments.

关 键 词：Gabor特征量 核函数判决分析 人脸识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]