Face recognition based on sparse representation classification often extracts the entire features, such as Eigenfaces, Randomfaces and Fisherfaces, but neglects the superiority of local features in terms of overcoming illumination and facial expression changes.For solving the above problems, the sparse representation for face recognition algorithm based on partitioning energy sub-band of Gabor wavelet is proposed.Firstly, face image is transformed by Gabor wavelet at different scales and orientations, every sub-band is divided into several blocks.Secondly, every sub-band is blend into eigenvector and each sub-band eigenvector is combined together to get enhanced Gabor eigenvector.Finally, such eigenvector is applied to sparse representation for face recognition.The result of the experiment indicates that the algorithm has more strongly robust to the change of illumination and facial expression.%基于稀疏表示分类的人脸识别通常提取特征脸、随机脸和费歇尔脸这些整体特征,忽略了局部特征在克服光照和表情变化方面的优越性.针对以上问题,本文提出了基于Gabor小波能量子带分块的稀疏表示人脸识别算法.首先将人脸图像进行不同尺度和方向下的Gabor小波变换,对得到的每个能量子带进行分块,然后将各子块能量信息融合组成子带的特征向量,再将各能量子带特征向量融合组成增强的Gabor特征向量,最后将该特征应用于稀疏表示人脸识别.实验结果表明,该算法对于光照和表情变化具较好的的鲁棒性.
展开▼
