Research On Relevance Feedback For Content Based Image Retrieval

While people benefit a lot from the advent of information digitized technology, they also have to face the dreadful dream about how to analyze, store and retrieve the huge amount of data efficiently and effectively, especially for those multimedia data. This paper focuses on the relevance feedback techniques in content based image retrieval and try to make this paper helpful for research and application of content-based image retrieval.This paper studies the relevance feedback technique of content based image retrieval from the following points of view:(1) Research on color feature extraction. A new variance analyzing based color quantization method is studied. The method is compared with other five quantization methods in terms of the root mean squared error. Experimental results show that the variance based color quantization method produces results that are far superior to other popular image quantization algorithms. Color coherent histogram is used for color feature extraction;it combines Color and spatial information.(2) This paper proposed a relevance feedback approach where the weights are the ratios of standard deviations of the feature values both for the whole database and also among the images selected as relevant by the user. It incorporates the human visual perception in the retrieval procedure. It optimized the retrieval result.(3) Research on Bayesian decision theory based relevance feedback mechanism. It is a kind of query shifting method. Experimental results show that the method is an effective relevance feedback method.(4) This paper designs a content based image retrieval (CBIR) system with relevance feedback mechanism as the test bed for retrieval algorithms, which is an experimental frame system. The CBIR system runs well with a largeimage library. In paper also summarized the promising research directions ofcontent based image retrieval system.