Illumination Invariant Face Recognition Using Gabor Wavelet Function

Abstract

Illumination variation changes the appearance of faces and makes it very difficult for accurate recognition. Face identification in an uncontrolled situation is still a challenging task for the researcher. In this paper, we propose a Fourier transform (FT) based illumination normalization and Gabor wavelets based feature extraction for perfect face representation for better classification. Implementing FT and Gabor wavelets in this method is to make the system more robust to the various constraints like illumination and noise where the performances of the other systems are degraded. Considering the phase magnitude information in the frequency domain, the illumination is compensated and using the Gabor filters noise and other unwanted disturbances like facial expressions are discarded. The extensive experimental results, on the publicly available Extended Yale-B face database, show that the proposed method outperforms the well-known face recognition methods even if on the extremely poor illumination images.