Performance Analysis of Light Illuminations and Image Quality Variations and its Effects on Face Recognition

Abstract

Nowadays face recognition is considered as an active and important area of research in the field of biometric technology as it is a non-contact method. Face recognition technology is employed in wide area of applications such as access management, authentication and also used in defence surveillance systems. The challenges like quality of image, different light illuminations, pose of a person and distance of person from camera may affect accuracy of face recognition. To overcome some of these challenges we proposed and evaluated our approach with two preprocessing techniques for image enhancement namely Histogram equalization (HE) and Contrast limited adaptive histogram equalization (CLAHE). The results are validated with variations in light intensity and person to camera distances in indoor as well as outdoor environments. The performance is measured in terms of Peak signal-to-noise ratio, mean square error and entropy and is evaluated for distance variation from 1 meter to 100 meter and lux variation from almost 0 lux to 32000 lux.