Color Image Compression with a Variable Block-Size Transform Scheme

Book excerpt: In a previous paper an adaptive transform-based coding method was introduced and used for the particular application of compressing the monochrome images of video conference sequences [1]. In the present paper the results of applying this adaptive scheme to compress still color images are presented. The basic principle of the algorithm is to encode those larger 2N x 2N -pixel regions featuring low image activity with a single 2N- point transform, and code high activity 2N x 2N- pixel regions with four N- point transform operations. Regarding the implementation issues, a strategy to reduce the computational complexity of the algorithm is also discussed.