Dissertation: Segmentation-based Coding of Monochrome and Color Images - In this research, images are approximated as sets of regions, and encoded as sets of region boundaries and region interior descriptions. The segmentation method used combines a Gaussian texture model with a Gibbs-Markov boundary model to find these regions with exceptional accuracy. An advantage of this approach is that it avoids the introduction of artificially structured artifacts, a common shortcoming of conventional transform coding methods. Applications include image compression, successive image transmission, image database indexing, and image enhancement.