基于Gabor小波和稀疏表示的人脸表情识别

Facial Expression Recognition Based on Gabor Wavelet and Sparse Representation

通过分析Gabor小波和稀疏表示的生物学背景和数学特性,提出一种基于Gabor小波和稀疏表示的人脸表情识别方法。采用Gabor小波变换对表情图像进行特征提取,建立训练样本Gabor特征的超完备字典,通过稀疏表示模型优化人脸表情图像的特征向量,利用融合识别方法进行多分类器融合识别分类。实验结果表明,该方法能够有效提取表情图像的特征信息,提高表情识别率。

By analyzing the biology background and mathematical properties of Gabor wavelet and sparse representation, a new approach for facial expression recognition based on Gabor wavelet and sparse representation is presented in this paper. Gabor wavelet transformation is adopted to extract features for the static facial expression image. The over-complete dictionary is constructed by the Gabor features of all training samples and sparse feature vector of this facial expression image is obtained by using sparse representation model. It uses a fusion recognition method for implementing multiple classifiers fusion. Experimental results show that integrating Gabor wavelet transformation and sparse representation is more effective for extracting expression image information. The approach effectively raises the accuracy of expression recognition.