Abstract: In order to improve the rate of face recognition under illumination change, a method of face recognition based on Local Contrast Enhancement（LCE） and Local Phase Quantification（LPQ） is put forward. The face image with uneven illumination is compensated by using facial symmetry combined with LCE algorithm. The enhanced images are marked with LPQ operators, and their dimensions are reduced with block Discrete Cosine Transform（DCT）. LPQ histogram sequence is calculated block by block as the description vector of face image, and it is recognized by the nearest neighbor classifier. The method is proved to be effectiveness with the experiment on the Yale B and CAS_PEAL database.

Key words: local contrast enhancement, local phase quantization, discrete cosine transform, face recognition

摘要： 为提高光照变化下的人脸识别率，提出了一种基于局部对比增强（LCE）和局部相位量化（LPQ）的人脸识别方法。采用面部对称的思想结合LCE算法对受不均匀光照的人脸图像进行光照补偿；利用LPQ算子对增强后的图片进行标记，并用分块离散余弦变换（DCT）进行降维；分块计算LPQ直方图序列作为人脸图像的特征描述向量，送入最近邻分类器进行分类识别。通过Yale B和CAS_PEAL数据库上的实验，证实了所提方法的有效性。

关键词: 局部对比增强, 局部相位量化, 离散余弦变换, 人脸识别