Fuzzy adaptive Voronoi-diagram-based clustering of textured images
Antony T. Popov, Edward R. Dougherty · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1998
Clustering means grouping of similar objects by optimizing a certain criterion. Clustering techniques are very common and useful in image processing and pattern recognition, especially for unsupervised initial segmentation of images. Images always contain some imprecise, uncertain or ambiguous regions and structures. Fuzzy sets constituted a good framework to represent and take into account these imprecisions. Their use leads to image processing methods where binary decision are not taken at intermediate levels, where only partial information is available. As mentioned by Bloch the usage of fuzzy sets in image processing relies on two different approaches - symbolic, where decision rules are derived from fuzzy logic, and numerical, where fuzzy sets directly represent the spatial structure of the image. Our work combines these two approaches.