Analysis of color image compression by BPG coder

Abstract

Lossy image compression is a popular way to get higher compression ratio, it also has several peculiarities if one deals with compressing images corrupted by noise. First, a specific noise filtering effect is observed. Second, optimal operational point (OOP) might exist where quality of a compressed image is closer to the corresponding noise-free image according to a chosen quality metric. In this case, it is worth compressing this image in OOP area. These peculiarities have been earlier mainly studied for grayscale images. In this paper, we analyze compression of color images corrupted by additive white Gaussian noise using better portable graphics (BPG) encoder in the cases of different chroma subsampling. Based on simulation results obtained for a set of color images, the initial recommendations on encoder parameter setting are given.