SEAN: A simple expanding-tree algorithm based on mean-division for color quantization

Abstract

The color quantization is an important operation in determining a set of K representative colors that resemble the N colors in an image. This investigation is motivated by the desire to demonstrate the fast, high quality reproduction of color image with a qualified color palette. This work develops an algorithm for a color palette, which firstly utilizes an expanding tree and splits the node with the greatest error distortion into eight children nodes, and then fine-tunes all terminal leaves to approximate its nearest cluster. Experimental results reveal that this algorithm outperforms Octree, Median Cut, and HF approaches.