知识与财富的链接
提出了一种基于Gabor小波变换和监督等距映射(supervised isometric feature mapping,S-ISOMAP)的人脸识别方法.针对流形学习算法不能消除图像特征向量中高阶相关信息的缺点,引入Gabor对归一化的人脸图像进行多方向、多分辨率滤波,并提取其对应不同方向、不同尺度的多个Gabor幅值特征(Gabor magnitude feature,GMF),然后使用具有提取鉴别子流形的S-ISOMAP算法对GMF特征进行维数约简,最后使用最近邻分类器进行分类.该方法综合运用了Gabor特征对人脸图像的优异的表征能力、S-ISOMAP的非线性维数约简能力,使得该方法对光照和表情变化等具有良好的鲁棒性.在YaleB和PIE人脸库上的实验表明了该方法的有效性.
In this paper, a method is proposed to recognize the face using Gabor feature and supervised isometric feature mapping(S-ISOMAP). Since the original feature vectors may include redundancy such as high-order correlation which cannot be removed by manifold learning algorithms, Gabor wavelet is introduced as a method to extract their corresponding Gabor magnitude features (GMFs) by convolving the normalized face image with multi-scale andmulti-orientation Gabor filters. Then, S-ISOMAP operates on GMFs to extra...
