Gray hausdorff distance measures for comparing face images

Human face recognition is considered to be one of the toughest problems in pattern recognition domain. The variations in face images due to differing expressions, pose and illumination are the key issues to be addressed by a face recognition system. In this paper a new measure, namely gray Hausdorff distance, is proposed to compare the gray images of faces directly while the existing Hausdorff distance-based measures are defined for binary images. An efficient computation of this measure in linear time is presented. The performance of this measure is evaluated on benchmark face databases for face recognition. The recognition system based on the new measure is found to be robust to considerable pose and expression variations as well as slight illumination variation. Comparison studies show that the proposed system performs better than the existing ones in many cases. Copyright © IICAI 2005.