Fuzzy Clustering of Color Textures using Skew Divergence and Compact Histograms: Segmenting Thin Rock Sections

Digital image segmentation is the process of assigning distinct labels to different objects in a digital image, and clustering techniques can be used to achieve such segmentations. However, many traditional segmentation algorithm fail to segment objects that are characterized by textures whose patterns cannot be successfully described by simple statistics computed over a very restricted area. In this paper we present a fuzzy clustering algorithm that achieves the segmentation of images with color textures by employing a distance function based on the Skew Divergence, that is based on the well-known Kullback-Leibler Divergence. In order for such a distance to produce good results when applied to color images, we reduced the dimensionality of the image's histogram, thus eliminating the sparsity of the color histogram and speeding up the execution of the algorithm. We performed experiments on thin rock sections and compared our results to the segmentations obtained by the Fuzzy C-Means and by another fuzzy segmentation technique, showing the superiority of our approach