Face Recognition Using LBPH Descriptor and Convolution Neural Network
V. Betcy Thanga Shoba, I. Shatheesh Sam · 2018
Face Recognition is an exigent problem in Biometrics and Computer Vision. It has become a wonderful field for researchers and can be widely used in applications like surveillance and security also. To create strong and distinct features, increase the inter-personal variations and decrease the intra-personal variations simultaneously remains a demanding problem in facial recognition. In this paper, the researcher explains how to improve the ability of face recognition system using Local Binary Pattern (LBP) for feature extraction and Convolution Neural Network (CNN) for classification of the images. The correspondence between the trained images helps CNN to converge faster and achieve better accuracy. There is a great improvement compared to other traditional methods too. To evaluate the accomplishment of this new method, it is found that higher face recognition accuracy can be achieved with less computational cost. The proposed framework is tested on the Yale dataset and achieved an accuracy of 98.6%.