The Clustering Algorithm Based On Finite Mixture Model And It's Application

Finite mixture model is a flexible and powerful tool for analyzing complicated phenomena, which provide an efficient method of simulating complicated density with simple structures and present a natural frame and semi-parameter structure of modeling unobserved population homogeneity and heterogeneity.Clustering based on finite mixture models is a kind of important clustering analysis methods, and the EM algorithm(Expectation-Maximization algorithm) is an important method of parameter estimation of mixture models. The traditional EM algorithm is sensitive to initial clustering center, and therefore how to choose initial values has become an important problem in realizing clustering based on finite mixture models using the EM algorithm. In this paper, a clustering algorithm based on grid to initialize the EM algorithm has been put forward that aims at improving EM algorithm of initial sensitivity and enabling it to achieve a better clustering effect. This algorithm identifies outlier and noise points based on the densities of grid cells, uses similarity measure to cluster analysis, and takes advantage of the idea of grid cores to lower time complexity. It request for scanning data set only once. Simulation results show that this method has low time complexity, and after optimizing the initial clustering center with the method the EM algorithm has good stability and precision.How to choice the optimum number of components in the mixture is an important but very difficult problem in the clustering based on finite mixture model.Many of previous works on this problem are reviewed in this thesis, including the methods based Bayesian theory and information/code criteria. And we emphasize the MML-EM algorithm.An image retrieval method based on Gaussian mixture models clustering is put forward through the research on clustering algorithms applied in image processing. The retrieval method extracts each image's features firstly, and establish Gaussian mixture models in data set with the eigenvalues. Thereby get Gaussian mixture models of all images. Then Clustering the data sets with Gaussian mixture models set of all images using the algorithm based on Gaussian mixture models. Output the class containing Sample image, namely get the retrieval results.