Using Color Variation to Mimic Human Visual Perception for Color Image Quantization

Color quantization is an important part of image processing, image formation, and related research. This paper presents a novel color image quantization algorithm which addresses the question of how to incorporate principles of human perception of color variation into the process of quantization. The color variation measure, calculated first in CIE Lab color space, is used to evaluate color variation and to coarsely segment the image. Considering both color variation and homogeneity of the image, the number of colors that should be used for each segmented region can be determined. Finally, the CF-tree algorithm is applied to classify pixels into their corresponding palette colors. The quantized error of the proposed algorithm is small because it considers color variation and because it is based on a model of human visual perception. Experimental results suggest that the proposed algorithm is more effective in quantizing color images than a wide variety of previously proposed algorithms.