Fuzzy VQ algorithms for color quantization

Two new extensions of Fuzzy C-means (FCM) algorithm which minimize an objective function incorporating a validity index are proposed. These algorithms are applied to color quantization of images. In the first approach, we minimize an objective function including a term for partition index. This algorithm attempts to place the cluster centers such that the membership values of the pixels are maximized. In the second approach, we minimize an objective function including an inter-cluster separation term. The goal here is to move cluster centers apart from each other towards the convex hull of the color space, hence obtaining a color palette which is more suitable for dithering, an operation generally applied after the quantization of the images.