Face Recognition by SVM Using Local Binary Patterns

Abstract

Authentication of the objects of interest plays a vital role and applicability in security sensitive environments. With Pattern recognition to classify patterns based on prior knowledge or on statistical information extracted from the patterns provides various solutions for recognizing and authenticating the identity of objects or persons. Identifying faces/objects of interest requires taking samples for training the classifier and classifying the input probe images with better recognition rate depending on the classification features. Facial recognition accuracy decreases when illumination of image is changed and with Single Sample per Person, where only one training sample is available does not give best matching results. In this paper, we present a model which works by taking different sample images and extracting Local Binary patterns, constructing the normalized histograms for training the SVM classifier and then classifying input probe images using Binary and Multiclass Support Vector Machines.