Research On Clustering Based On Partition And Its Application To Image Segmentation

Clustering is a grouping process by which similar objects are in same group whereas disimilar objects are in dfiffertent group. Up now, there are many clustering algorithms in references which are classified as five kinds of type,namely partitioned clustering approach,hierachied clustering approach,grid clustering approach,density clustering approach and model clustering approach. In these different clustering algorithms, partitioned clustering is most common method.In this paper, we study the partitioned clustering method and its application to image segmentation. It consists in three aspects contents as follows:By analyzing and studying clustering algorithms based on entropy, we give a unify model based entropy clustering algorithm and present a based unified model fuzzy clustering algorithm. Moreover, joining with kernel trick, we also present kerneled fuzzy clustering algorithm based entropy.Aimming at image segmentation and image's characteristic, we present image segmentation method and kerneled fuzzy clustering image segmentation methods.As there exist a large number data in image segmentation, by introducing data reduction, we study image segmentation based on C means clustering algorithm.