A Comparative Study of Face Authentication Using Extreme Learning Machine, Euclidean and Mahalanobis Distance Classification Methods

Face recognition can be used for both verification and identification. Today face recognition technology is being used to combat passport fraud, support law enforcement, identify missing children, and minimize benefit or identity fraud. The two main steps in a face recognition system are: (i) to define an effective representation of the face images, which includes sufficient information of the face for future classification, (ii) to classify a new face image with the chosen representation. In this paper, Extreme Learning Machine method for face recognition is proposed and compared with Euclidean and Mahalanobis distance methods for better face recognition rate. The Mahalanobis distance is a metric which is better adapted than the usual Euclidean distance to settings involving non spherically symmetric distribution, where as extreme learning machine (ELM) is an efficient learning algorithm for generalized single hidden layer feed forward networks (SLFNs), which performs well in classification applications. This will further enhance the quality of facial image authentication. Various experiments are done for 400 samples from ORL database for the three methods and the results are analyzed.