Abstract

The ever-increasing requirements of security concerns have placed a greater demand for face recognition surveillance systems. However, most current face recognition techniques are not quite robust with respect to factors such as variable illumination, facial expression and detail, and noise in images. In this paper, we demonstrate that face recognition using support vector machines are sufficiently robust to different kinds of noise, does not require image pre-processing, and can be used with rather low image resolutions.

Authors

Prashanth Harshangi¹, Koshy George²
PES Centre for Intelligent Systems, India¹, PES Institute of Technology, India²

Keywords

Face Recognition, Support Vector Machines