Constrained and unconstrained simplification of image partitions encoded with the method of transition points

The encoding of image partitions plays a key role in the overall efficiency of the so-called region-based image coding techniques. A method based on the novel concept of "transition point" was previously proposed for the lossless encoding of arbitrary contour maps. This paper addresses the problem of partition simplification in the context of this new encoding method. A new simplification approach is introduced and results concerning its compression efficiency under constrained and unconstrained conditions are presented.