Abstract:

In order to improve the face recognition rate under varying lighting conditions,a novel face recognition algorithm based on the nonsubsampled Contourlet transform and the pseudo-Zernike moment is proposed. In this algorithm,first,invariant illumination components are extracted via the soft-threshold denoising in the Lambertian illumination model. Then,the corresponding pseudo-Zernike moment vectors are calculated and are used as face classification features. Experimental results on Extended YaleB and CMU PIE face databases show that,as compared with the common face recognition algorithms,the proposed algorithm can eliminate the effect of illumination more effectively and adapt to the variation of scale and pose,so that it significantly improves the accuracy of face recognition.

Key words: face recognition, nonsubsampled Contourlet transform, pseudo-Zernike moment, illumination model