Fuzzy-based illumination normalization for face recognition

In this paper, we address the problem of reducing the effect of illumination especially for human face recognition. We create an adaptive contrast ratio based on Fuzzy by considering two models of individual face as input, appearance estimation model and shadow coefficient model. We then apply a Genetic Algorithm to optimize the Fuzzy's rule. Principal Component Analysis (PCA) and Nearest Neighbor (NN) based on correlation distance are used as the classifiers. We test our algorithm for both still image and natural scene video to show its feasibility for real time system. The experimental results are also provided to prove the robustness and performance of our algorithm in order to recognize desired person under variable lighting conditions.