S3TC Image Compression using Clustering

A.S. Aguado and E. Montiel (UK)

Keywords

Image compression, Image processing, Clustering,Principal Component Analysis, DXT compression.

Abstract

S3TC or DXT is a lossy image compression format widely used in graphic applications. Many applications use this format to store images representing material properties such as texture colours, surface normals as well as rendering information such as pre-computed lighting and environment maps and even animation information. Its main advantages are that it facilitates random access to pixels and that it can be fast decompressed since it is widely supported by graphics hardware. However, S3TC does not define an optimal compression algorithm and it is possible to develop several compression techniques that compromise quality and processing complexity. This paper evaluates several compression strategies and it presents a new compression method that includes clustering for colour quantization. The method is not dependent on optimizations of particular hardware architectures and experimental results show that it can effectively reduce compression errors without excessive computational requirements.

Important Links:



Go Back