SIFT & MRLBP Descriptors Based Age Invariant Face Recognition

Abstract

Automatic face recognition is an important problem, but age invariant face recognition is a major challenge. The face appearance of a person is subject to significant change due to age progression over time. In this paper, the discriminative model is proposed to match face images of a subject at different ages. To develop a discriminative model for age invariant face recognition based on an appropriate feature representation and classification. Where Local Feature Description is used for feature representation and classification is done using MFDA. In this approach, each face is represented by designing a local feature description scheme. It consists of Scale Invariant Feature Transform (SIFT) and Multi-scale Robust Local Binary Patterns (MRLBP) which serve as local descriptors. In MFDA, multiple LDA-based classifiers are constructed and these classifiers are combined to generate a robust decision by using a fusion rule. The superiority of proposed method is examined and demonstrated through FGNET database. For the discriminative model with SIFT and MRLBP, the recognition accuracy is 92.38%.