Face recognition using Elastic bunch graph matching

Abstract

A closed-set identification is implemented using Elastic bunch graph matching (EBGM) algorithm. It uses cosine similarity as its matching criterion instead of a classifier for recognition. The proposed method makes use of facial features like fuducial points to differentiate between faces. It is insensitive to variation in facial expressions, illumination and poses on frontal and ¾ frontal images. Experimental results show that the proposed method can achieve a recognition accuracy of 96.67% for the training to test ratio of 7:3 on face images. This method can be extended to provide profile face recognition.