Abstract
The optimization of a monochrome still picture coding system is discussed. The coding scheme is based on a two-component model: a structural one (plateau component) and a texture component. The coding uses first a segmentation process based on edge-detection and contour-following technique, which produces connected and closed homogeneous regions. The picture associated with the plateau component is optimally coded and decoded by special techniques used in image synthesis. The texture component is optimally coded by an adaptive discrete cosine transform coding scheme after some specific preprocessing. The desired compression ratio is high (<1 bit/pel on a typical TV scene) while conserving an excellent visual quality.

This publication has 3 references indexed in Scilit: