An Improved Illumination Invariant Face Recognition Based on Gabor Wavelet Transform

Abstract

Face recognition under unconstrained surroundings have been a demanding task due to various constraints such as expression, illumination, pose, occlusion etc. Over the years, researchers have well recognized the deployment of wavelet transform for extracting robust features from facial images under such constraints. An improved illumination invariant approach based on Gabor wavelet transform (GWT) is presented for face recognition. Firstly, Gamma intensity correction (GIC) is applied to correct the total intensity of database images. Secondly, from these images, robust features are extracted using Gabor wavelet transform. Dimensions of these extracted features are reduced by applying dimensionality reduction approach such as principal component analysis (PCA). Lastly, K-nearest Neighbor approach is used for classification. Experimental result on Yale dataset indicates that the proposed approach enhances the recognition accuracy efficiently in contrast to other existing techniques, achieving recognition accuracy of 100%.