中文摘要：

为了提高用低分辨率图像的人脸识别的效果，提出了一种基于完备局部相位量化（CLPQ）的低分辨率人脸识别方法。该方法分别用CLPQ—P算子和CLPQ—M算子提取低分辨率人脸图像的相位特征和幅值特征以形成CLPQ直方图作为图像的特征向量，然后进行直方图相似性比较，最后通过最近邻分类器进行人脸识别。实验表明，该算法对ORL和Yale低分辨率人脸图像的识别效果出．LPO方法更女千．鲁椿忡再高．

英文摘要：

To effectively realize the face recognition from low resolution images, a face recognition method based on com- pleted local phase quantization （CLPQ） is presented. The new method uses the operators of CLPQ _ P and CLPQ _ M to extract the phase features and the magnitude features from low resolution face images respectively, so as to form the CLPQ histogram as the feature vector of face images. Then, it performs the similarity comparison for the histogram , and finally, uses a nearest neighbor classifier to achieve face recognition. The results of the experiments on the low resolution face images from the ORL and Yale databases illustrate that the presented method has the bet- ter recognition rate and robustness than the LPQ method.