组合局部多通道Gabor滤波器和ICA的人脸描述与识别

Method for face description and recognition by local multi-channel Gabor filters and ICA

针对人脸识别是当前人工智能和模式识别的研究热点,提出了一种组合局部Gabor滤波器组和ICA技术(简称LMGICA)的人脸描述方法,首先对归一化的人脸图像进行采样分块,然后对局部子块进行多方向、多分辨率Gabor小波滤波,并提取其对应不同方向、不同尺度的多个Gabor幅值域图谱(local Gabor magnitude map,LGMM),接着由滤波图像直接构建高维特征矢量;再将这些高维特征矢量通过主成分分析进行降维;最后采用ICA技术分析和提取降维后的特征矢量中的独立成分用于识别分类。通过与经典Gabor滤波器、PCA、ICA等方法的对比实验,验证和评价了本方法的性能。

Face recognition is an active research area in the artificial intelligence. This paper proposed a new face recognition algorithm using the RBF network based on local multi-channel Gabor filters and fixed point ICA （LMGICA）. The normalized face image was firstly sampled and blocked, and then the blocked face image was filtered by multi-orientation Gabor filters with multi-scale to extract their corresponding local Gabor magnitude map （LGMM）, which were constructed to higher dimensional feature vectors. Next, reduced the dimeusionality of these vectors by means of principal component analysis （PCA）. Finally, analyzed and extracted the independent components in the resulting vectors with dimensionality were reduced by using ICA recognition classification. The experiment results demonstrate and evaluate performance of the method compared by the classic Gabor filters and PCA, ICA.