Analysis of visual quality for denoised images
Andrii Rubel, Oleksii S. Rubel, Volodymyr Lukin · 2017
This paper presents results of our study intended on analysis of visual quality of different types of images subject to denoising. More than 100 experiments with observers have been carried out to assess visual quality of denoised images. Two filters - standard DCT in sliding blocks and known BM3D filter - are considered. The latter filter often provides better visual quality but not sufficiently. It is shown that effect of denoising is not always positive. There are quite many practical situations when denoising is practically useless and/or not worth applying.