基于NSCT和SQI的光照不变量及人脸识别

Illumination Invariant and Face Recognition Based on NSCT and SQI

为了消除光照变化对人脸识别的影响,提出了一种结合无下采样轮廓波变换(Nonsub-sampled contourlet transform,NSCT)和自商图像(Self-quotient image,SQI))的光照不变量提取算法。该算法首先对图像进行伽玛校正,一定程度上减弱了不同光照条件的影响,然后使用NSCT对伽玛校正后图像进行多尺度多方向分析,使用自适应NormalShrink方法对各高频子带进行滤波,通过反无下采样轮廓波变换得到平滑图像,最后利用SQI提取光照不变量。在YaleB与CMUPIE人脸库上的实验结果表明:所提出的算法能够有效消除光照变化对人脸识别的影响,识别率高于多尺度Retinex(Multiscale Retinex,MSR)、SQI和对数全变差(Logarithmic totalvariation,LTV)等方法。

In order to eliminate the effect of varying illumination on face recognition,a novel illumination invariant method based on nonsubsampled contourlet transform（NSCT） and self-quotient image（SQI） is proposed.The method first performs Gamma correction on image under various lighting conditions,which can decrease the effect of varying illumination to some extent.The NSCT is used for multiresolution analysis.NormalShrink filtering is applied to high frequency subbands and a smooth image can be obtained by inverse nonsubsampled contourlet transform.A self-quotient image is used for illumination invariant extraction.Experimental results from Yale B and CMU PIE databases show that the proposed method can effectively eliminate the effect of varying illumination on face recognition,and the recognition rate of the method here is higher than the multiscale Retinex,self-quotient image and logarithmic total variation methods.