Gamma Correction Technique Based Feature Extraction for Face Recognition System

One of the most important challenges for practical face recognition systems is to make recognition more reliable under uncontrolled lighting conditions. We tackle this by using novel illumination-insensitive preprocessing method. The proposed face recognition system consists of a gamma correction, a preprocessing stage, a hybrid Fourier-based facial feature extraction, and Principal Component Linear Discriminant Analysis (PCLDA). Gamma Correction is a nonlinear gray-level transformation that replaces gray-level I with Iγ (for γ > 0), where γ is a user-defined parameter. In the preprocessing stage, an “Integral Normalized Gradient Image”, (INGI) is obtained by transform a face image into an illumination-insensitive image. The effect of illumination gets reduced in the INGI by normalizing and integrating the smoothed gradients of a facial image. Using frequency band model selection the hybrid Fourier features are extracted from three different Fourier domains in different frequency bandwidths and further by adding PCLDA the robustness of the system gets improved. In face recognition, it is not possible to process with the entire extracted features, hence the dimension of the feature vectors has to be reduced. In this paper, this is done by using the linear method called PCLDA. The proposed system using the Yale B data set which is having a 2-D face images under various environmental variations such as illumination changes and expression changes. Keywords-Gamma Correction, Preprocessing, Feature extraction, PCLDA and Yale B data set