Shape feature based extraction for face recognition

In this paper, we present a feature-based approach to fast face recognition. A novel shape based automatic reference landmark and feature extraction technique is proposed for face representation. The difference between two faces is measured by a set of the extracted features and three-dimensional (3-D) features are used for a face template registration. Unlike holistic face recognition algorithms, the feature-based algorithm is relatively robust to face variation of expressions, illumination and pose, due to invariance of its facial feature vector. The proposed approach achieved promising identification rate and verification rate on a subset of Massachusetts Institute of Technology-CBCL face recognition database.