Human Face Recognition Using Combination of ZCA Feature Extraction method and Deep Neural Network

Nowadays, there are deep learning-based face recognition systems with reliable accuracy rates. But these systems need high computational powers because they should do feature extraction. This research presents the consequences of recognition rates on the AT&T face database, using the hybrid methods of ZCA feature extraction and ANN in order to reduce computational power usage.

I. Introduction

Face recognition is the process of identifying people by matching query face image with the template face image in the facial database due to the calculation of their pixel data. Face recognition is widely used in the biometric field because of its benefits such as low cost and non-intrusive nature. Face recognition has been started in psychology and engineering works of literature in the 1950s and 1960s. After the seminal work of Kanade, automatic face recognition systems have been started since the 1970s [1] . Thereafter, a lot of researchers have been explored many approaches to handle this task. But it still needs to reach human performance. Nowadays, there are a lot of methods of face recognition systems with reliable accuracy rates are used in various purposes, such as machine learning, computer vision and pattern recognition. There are many face recognition-based applications which exist in our sophisticated environment. The most recently used face recognition-based applications can be found in the surveillance system, access control, credit-card verification, mug shot searching, criminal identification and security of computer systems [2] .