Recognition of faces using discriminative features of LBP and HOG descriptor in varying environment

Recognizing faces in presence of illuminations, pose, facial expression variations in controlled as well as uncontrolled environments remains one of the most challenging aspect. In this paper, we propose a novel recognition methodology which deals with challenges of face recognition to obtain robust and efficient recognition. The framework is based on extracting discriminant statistical features from Local Binary Pattern and providing it to modified HOG descriptor after normalization. LBP-HOG feature vectors are used as an input to various classifiers. Discriminant analysis and distance based classifiers have been used to classify face images. The proposed method is systematically examined on several databases. Extensive experiments illustrate that feature vectors obtained from proposed algorithm are effective and efficient for dealing with challenges of face recognition. Experimental results evaluated using four databases illustrates the benefits of our approach.