结合Gabor变换和FastICA的人脸表情识别方法

提出了一种结合Gabor变换和FastICA技术的人脸表情特征提取方法。Gabor小波具有很好的空频局部性和多方向选择性，因此更有利于表情细节信息的提取。FastICA技术能够消除信号间的高阶统计冗余。对图像进行Gabor变换，把得到的系数排列成Gabor特征矢量，用FastICA对Gabor特征矢量进行特征提取，用K-近邻分类器进行分类。JAFFE表情库中的实验证明该方法的有效性。

Facial expression recognition based on Gabor transform and FastICA

An effective method for the facial expression feature extraction is presented by combining the Gabor transform with the Fast Independent Component Analysis（FastICA）.Gabor wavelets exhibit strong characteristics of spatial locality and orienta-tion selectivity,which are good for the extraction of the image’s texture.FastICA can reduce the redundancy of high-order statis-tics.The Gabor transform is carried out on each original image,and the outputs are concatenated into a Gabor feature vector.Fas-tICA approach is used to extract features from the Gabor feature vectors of all the images.The K-neighbor method is used for classification.A series of experiments performed on the JAFFE database indicate the efficiency of the proposed method.