Methodologies in Face Recognition for Surveillance

Face recognition being the most significant part of the biometrics and video surveillance systems has to deal with many challenges like pose, illumination, distance, expression variations along with occlusion, low resolution and noise. This paper analyses the latest approaches and methodologies adapted by Face Recognition systems in overcoming the challenges with the focus on their performance on publicly available bench-mark face databases. Few of the approaches discussed in this paper are Local Binary Pattern technique, Reference Face Graph framework, Principal Component Analysis algorithm, Linear Discriminant Analysis algorithm, MLP Neural Networks, 3D Modeling, Back Propagation Neural Network, Local Pavithra S G Bhat Gradient Hexa Pattern descriptor, FPGA based architecture, Conditional Generative Adversarial Networks and Trunk-Branch Ensemble Convolutional Neural Networks. This paper provides an insight for selection of an approach for future researchers and practitioners for real world implementation.