Efficient face recognition with compensation for aging variations

Face recognition is identifying a person based on facial characteristics. Automated face recognition is identifying a given query face called probe from a target population known as gallery. The face recognition algorithms perform well when the interpersonal images have more discriminating features than intra personal images. The changes in the face bring down the similarity of the intrapersonal images. The variations in the face can be due to pose, expression, illumination changes and aging of a person. Face recognition accuracy is largely influenced by the age related changes in face. Aging effects on face are not uniform and depend on both intrinsic as well as external factors like geographic location, race, food habits etc. The facial changes are exclusive for each person in spite of aging being an apparent phenomenon among all individuals. Hence there are many challenges still open in compensating age related variations. In this paper we have proposed a novel self-PCA based approach in order to consider distinctiveness of the effects of aging of a person for age invariant face recognition. The region around the eyes is used as the input feature instead of the entire face as it is more stable part of the face with respect to aging and also requires less space. The proposed approach is tested using the images of the FG-NET database.