VCF Lossy Compression of Images

Abstract

We explore the VCF lossy compression of images, based on visual degradations rather than mathematical transformations. We find a very simple algorithm that compares to JPEG compression rates, well-suited for real-time application (a few hundreds C lines, very low cpu usage, no convergence condition, very low memory requirements), with easier to control artefacts than most lossy algorithms. The simplicity of the algorithm allows the implementation of advanced user features, such as local compression constraints, application-specific compression algorithms (medical, fingerprints), animated pictures compression, and hardware implementation. An enhanced version of the VCF algorithm is proposed to make it have better compression rates than JPEG, while keeping the speed and simplicity.

Table of Contents

1. Introduction: the ideal image compression.
2. Basic VCF Algorithm.
3. Results for Grayscale Images.
4. Results for Black & White Images.
5. Future Work and Proposed Improvements.
6. Conclusion.

1. Download the test implementation (alpha3).
2. FAQ.