A Study of Contrast Enhancement Methods in Various Color Spaces on Biometric Recognition System

Abstract

Image enhancement especially contrast enhancement has a significant role in the biometric recognition system. Contrast image enhancement exposes an important feature in an image that is concealed around the dark area before the transformation. This brings important improvement in the system of biometric recognition. In this paper, we studied the effect of contrast enhancement methods such as histogram equalization, image adjustment, mathematical morphology in top-bottom hat algorithm and CLAHE in HSV, YCbCr and RGB color spaces. Using non-exhaustive cross-validation and nearest neighbor distance calculation, 400 images of androgenic hair were observed. The best results were obtained from saturation in HSV and Cr component from YCbCr using normal histogram equalization method. Overall, the contrast enhancement methods, in certain combination, improved precision recognition until 20% better. The best result in precision was derived from saturation component in HSV color spaces, around 89% precision recognition and Cr component in YCbCr color spaces, around 88% precision recognition. Meanwhile, the best contrast enhancement method was obtained from histogram equalization, around 89% precision recognition.