A Meta-Analysis of Contrast Measures Used for the Performance Evaluation of Histogram Equalization Based Image Enhancement Techniques
B. N. Anoop, P.E. Ameenudeen, Justine Joseph · 2018
Objective measures of image contrast are necessary to compare the performance of different enhancement schemes and to optimise their operational parameters. The metrics available in the literature meant for this purpose include Enhancement error (EME), `Contrast (C)', Standard Deviation (SD) and Edge Based Contrast Measure (EBCM). Such matrices should be well correlated with the variation in grey level contrast in the image and should be robust to other luminance factors. In this meta-analysis, the ability of the contrast metrics to account for image contrast is tested on a set of simulated images of known contrast, comprising of medical images as well as panoramic images. Among the contrast metrics, SD exhibited good correspondence with the variation in contrast and it is robust to other perturbing factors.