Segmentation of color map images based on color clustering and Markov models

摘要

This paper presents a method for segmentation of color map images. A clustering algorithm is used to coarsely segment the image into several perceptually uniform regions according to the observed feature vectors, then a knowledge-based selection procedure chooses the clusters of well-defined foreground regions. Finally, contextual classification based on Markov random field (MRF) is applied to the image for fine classification.