MEN: Multi-Attribute Feature Selection for Face Recognition

Abstract

Face recognition has become more sophisticated in biometric based authentication systems, which uses various facial features. The authentication mechanism is still required with more features to be used and has to be done in a short time. Existing face recognition algorithms are more scalable in time and memory used, also produces high frequency of false positive results. To overcome the problem of false positive results, we propose MEN-Mouth, Eye and Nose features based multi attribute feature selection method for face recognition. The proposed method extracts the facial features like Nose, mouth and eye, from extracted features we compute eccentric measures for each of the feature. The eccentric measure is computed between four axis co-ordinates of facial features. Computed features are converted into single feature, and computes feature weight based on computed feature set. The computed feature weight is used to recognize the person.