Review the strength of Gabor features for face recognition from the angle of its robustness to mis-alignment

Abstract

Gabor feature has been widely recognized as better representation for face recognition in terms of rank-1 recognition rate. In this paper, we review the strength of Gabor feature for face recognition from the new angle of its robustness to mis-alignment using a novel quantificational evaluation method combining both the alignment precision and the recognition accuracy. Our experiments show that, compared with the gray-level intensity, Gabor feature is much more robust to image variation caused by the imprecision of facial feature localization, which further support the feasibility of Gabor representation.