A new face recognition system based on color histogram matching

In this paper a new face recognition system based on the color histogram matching is proposed. The system uses color histograms of the extracted face regions as the feature vectors for recognition. The recognition is performed by identifying the subject that maximizes the cross correlation of the histogram of the given image and the histograms of the face subjects in the database. Majority voting among the decisions of the histogram based face recognition classifiers in different color channels have been combined to perform increased recognition rates. The proposed system reaches to 99.52 % recognition rate on the Essex University Database of 150 face subjects with varying illumination, background and pose.