A Novel Scheme to Improve Lossless Image Coders by Explicit Description of Generative Model Classes

TL;DR

This paper introduces a new lossless image compression scheme that explicitly describes generative model classes, achieving significant rate reductions by optimizing under a Bayesian framework.

Contribution

It proposes a systematic method to improve lossless image coders by explicitly modeling image classes and using Bayes codes for optimal compression.

Findings

Achieves approximately 19.7% reduction in average coding rates.

Validates the effectiveness of the model-based approach.

Enhances lossless image coding efficiency.

Abstract

In this study, we propose a novel scheme for systematic improvement of lossless image compression coders from the point of view of the universal codes in information theory. In the proposed scheme, we describe a generative model class of images as a stochastic model. Using the Bayes codes, we are able to construct a lossless image compression coder which is optimal under the Bayes criterion for a model class described appropriately. Since the compression coder is optimal for the assumed model class, we are able to focus on the expansion of the model class. To validate the efficiency of the proposed scheme, we construct a lossless image compression coder which achieves approximately 19.7% reduction of average coding rates of previous coders.