Visual experience of noise in digital images

Abstract

© 2016, Strojarski Facultet. All rights reserved. Noise in the image is always present, the only question is, to what extent it is visible to the human visual system. Some images could have a significantly high level of noise, but the human eye will not perceive that. The aim of this study was to investigate the influence of image content on the perception of image noise. We used three images with different contents (highlights, many details, and uniform surfaces with midtones). Samples were simulated by adding six levels of noise in MATLAB software, using the built-in function. Those samples were measured with image quality IQ metrics: SCIELab, SSIM and IQM2. For psychophysical scaling, we used the paired-comparison method (PC-index). The results obtained in this study showed that image content influenced the perception of image noise. The image with many details was the most difficult to estimate. Those paired samples looked the same, even though they had different levels of noise.