A fuzzy algorithm for color quantization of images

In this paper, we review a number of techniques for fuzzy color quantization. We show that the fuzzy membership paradigm is particularly suited to color quantization. where color cluster boundaries are not well defined. We propose a new fuzzy color quantization technique which incorporates a term for partition index. This algorithm produces better results than fuzzy C-means at a reduced computational cost. We test the results of the fuzzy algorithms using quality metrics which model the perception of the human Visual system and illustrate that substantial quality improvements are achieved. (C) 2002 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved.