GMM-ClusterForest: a novel indexing approach for multi-features based similarity search in high-dimensional spaces

摘要: This paper proposes a novel clustering based indexing approach called GMM-ClusterForest for supporting multi-features similarity search in high-dimensional spaces. We fit Gaussian Mixture Model (GMM) to data through the Expectation-Maximization (EM) algorithm estimating GMM parameters and Minimum Description Length (MDL) criterion selecting structure. Each component is taken as cluster center each point assigned according Bayesian decision rule. By performing this method hierarchically, an index tree constructed corresponding developed type of features. Then fulfilled by fusing trees all types features considered. evaluated proposed applying it example-based image retrieval conducting experiments on Corel 1000 dataset self-collected large dataset. The experimental results show that our effective promising.