Color Palette Extraction by Using Modified K-means Clustering

Color can profoundly influence the emotion of human visual perception. Extracting the color palette for describing colors in an image has extensive applications in many areas, such as graphic design, image classification and image recolorization. In this paper, we propose an improved method to extract the color palette based on k-means clustering and the analysis of hue and saturation components of the image colors. The experimental results show the improvement in saturation of representative colors and color extraction in small image detail. From the evaluation results, the respondents are satisfied in color palettes of the proposed algorithm for all types of tested images, i.e., photo, scan and computer graphic images.