摘要： 光照变化和环境噪声会引起人脸识别正确率下降。为此，提出一种基于Curvelet变换和LBP的可变光照人脸识别方法。对原始人脸图像进行Curvelet变换，对第1层低频系数，采用对数运算和局部二值模式运算克服光照影响，舍弃剩余的最高频信息子图像，以除去环境噪声和光照产生的阴影边界带有的虚假信息，利用主成分分析和线性判别分析方法进行人脸识别。仿真结果表明，该方法能有效去除光照和噪声引起的影响，具有较好的鲁棒性。

关键词: 人脸识别, 光照, 噪声, Curvelet变换, 局部二值模式, 主成分分析, 线性判别分析

Abstract: A variable illumination face recognition method based on Curvelet and Local Binary Pattern(LBP) is proposed to solve the recognition rate decline problem because of the variant illumination and noise. The original face images are decomposed by Curvelet transform. It processes the coefficients of the first layer by logarithm computing and LBP method to remove the effect of the illumination, and directly discards the surplus highest frequency information to alleviate the environment noise and the fake information at the intersection of the shadow. The face recognition is done by the Principal Component Analysis(PCA) and Linear Discriminant Analysis(LDA) using the preprocessed the first layer data and the other layer data. Simulation results show that this method can alleviate the effect of the illumination and environment noise, and has good robustness.

Key words: face recognition, illumination, noise, Curvelet transformation, Local Binary Pattern(LBP), Principal Component Analysis(PCA), Linear Discriminant Analysis(LDA)

中图分类号: 

• TP391