Color image quantization using distances between adjacent colors along the color axis with highest color variance

This paper describes a simple but effective hierarchically divisive colormap design technique for color image quantization. By sorting colors based on their components along the principal axis, the one with the highest variance of color distribution, the Euclidean distances between any adjacent colors' along the axis are used to find the cutting plane that is perpendicular to the axis and divides a color cell into two subcells with approximately equal quantization errors, with respect to their centroids. As a result, the total quantization error on both cells is minimal. The experimental results reveal that the proposed algorithm is effective and yields a better execution time when compared with others. Moreover the proposed method performs well on both pictures with 15-bit and 24-bit colors regardless of a number of colors in the colormap. (C) 2004 Elsevier B.V. All rights reserved.