Illumination invariant face recognition using MSVM

Abstract

In Computer Science, Face Recognition is a challenging research field and face Recognition with variation in lighting condition is an even more challenging and difficult task. Numerous researches has been proposed on various face recognition methods, however, there have not been any successful research on illumination invariant face recognition. Here, we propose an illumination invariant face recognition system using mSVM with RBF Kernel. The first necessary step required is face detection which we achieve by using a widely known method name Viola Jones algorithm. To overcome illumination problem Median Filter and Gabor Filter is proposed to be combined at a pre-processing stage which reduce image noise and contrast images are adjusted so accuracy of face recognition is increased. We also proposed that PCA method may be used for extracting feature vectors which increase accuracy of the proposed face recognition model.