Non-hierarchical Clustering with Rival Penalized Competitive Learning for Information Retrieval
In large content-based image database applications, efficient information retrieval depends heavily on good indexing structures of the extracted features. While indexing techniques for text retrieval are well understood, efficient and robust indexing methodology for image retrieval is still in its infancy. In this paper, we present a non-hierarchical clustering scheme for index generation using the Rival Penalized Competitive Learning (RPCL) algorithm. RPCL is a stochastic heuristic clustering method which provides good cluster center approximation and is computationally efficient. Using synthetic data as well as real data, we demonstrate the recall and precision performance measurement of nearest-neighbor feature retrieval based on the indexing structure generated by RPCL.
Unable to display preview. Download preview PDF.
- 4.T. C. Chiueh. “Content-Based Image Indexing”. In Proceedings of the 20th VLDB Conference, pages 582–593, September 1994.Google Scholar
- 8.I. King, L. Xu, and L.W. Chan. Using rival penalized competitive clustering for feature indexing in Hong Kong’s textile and fashion image database. In Proceedings to the International Joint Conference on Neural Networks (IJCNN‘98), pages 237–240. IEEE Computer Press, May 4–9, 1998.Google Scholar
- 10.R. Rosenblatt. Principles of Neurodynamics. Spartan Books, New York, 1959. neural networks.Google Scholar
- 11.T. Sellis, N. Roussopoulos, and C. Faloutsos. “The R+-tree: a dynamic index for multidimensional objects”. In Proceedings of the 13th VLDB Conference, pages 507–518, 1987.Google Scholar
- 13.P. N. Yianilos. “Data structures and algorithms for nearest neighbor search in general metric spaces”. In Proc. of the 4th Annual ACM-SIAM Symp. on Discrete Algorithms, pages 311–321, 1993.Google Scholar