Image Retrieval Using Mixture Models and EM Algorithm
This paper presents an original system for interactive content-based image retrieval (CBIR). A novel approach for searching by similarity is introduced. It is based on a classification of the index database using mixture models and the EM algorithm. The presented retrieval system is evaluated and validated using a medical image database and the Washington University heterogeneous database (ANN).
KeywordsMixture Model Image Retrieval Relevance Feedback Relevant Image Image Retrieval System
- 1.C. Ambroise, G. Govaert, “Em algorithm for partially known labels”. IFCS2000, 7th Conference of the International Federation of Classication Societies, Namur, Belgique, pp.11–14, 2000.Google Scholar
- 2.Boutry, N., Lard, A., Solau, E., Flipo, R., Cotten, A.: The Usefulness of MR Imaging in Early Rheumatoid Arthritis. RSNA, Chicago (2001) 25–30Google Scholar
- 7.S. Santini, R. Jain, “Similarity Measures”, IEEE T-PAMI, Vol. 21, n 9, 1999.Google Scholar
- 9.A. Smeulders, M. Worring, S. Santani, A. Gupta, R. Jain, “Content-Based Image Retrieval at the end of the Early Years”, IEEE transactions on pattern analysis and machine intelligence, Vol. 22, n12, 2000.Google Scholar
- 10.Y. Wu, Q. Tian, T. Huang, “Discriminant-EM Algorithm with Application to Image Retrieval”, Proc. Computer Vision and Pattern Recognition, pp. 222–227, 2000.Google Scholar