A Novel Method for Image Clustering

摘要: Image clustering has been attracting mounting focus on widely used fields, such as data compression, information retrieval, character recognition and so on, due to the emerging applications of various web-based mobile-based image re- trieval services. To study this, based Voronoi diagram, we propose a novel algorithm effective discovery clusters in this paper. More specifically, diagrams at first, number irregular grids are built across whole plane. Furthermore, leveraging good property "the nearest neighbor" for diagrams, plane assigned by points different clusters. On one hand, density grid points, it automatically adjusts final suitable clustering; other according changes centroids, tunes positions Voronoi's seeds. At last, cells finally become result process. The empirical experiment results show that our proposed method not only can cluster dataset effectively, but also achieve comparative performance with X-means K-means algorithm. Moreover, outperform effectiveness both DBSCAN OPTICS algorithms, which classic density-based algorithms towards larger- scale real-world applications.