Retrieving Image Resource Technique Based on Bayes Semantic Classification and Visual Feature Extraction

Abstract

An image retrieval method was proposed based on feature extracted in DCT and semantic classification by bayes. Firstly, Semantic need for retrieving digital image was studied, and Simple Bayesian Classifier was used in semantic classification of image resource; secondly, improved image feature of edge space distribution probability based on DCT was extracted to obtain edge information of goal object and to set up 20 Eigen values. Thirdly, indexing feature vectors were obtained through semantic classification for semantic filtration, by which the retrieval efficiency was improved. With experiment, image resource database is established and the proposed algorithm shows better results by test on precision comparison compared with other methods.