Studies On The Semantic Image Retrieval Based On Bayes Statistical Learning Theory

With swift development of technologies of computer and internetwork communication, as well as the widespread application of multimedia technologies, various kinds of magnanimous information are being gathered, transmitted and applied in forms of graph, image, animation and video, besides the traditional previous text mode. Image retrieval is drawing more and more attention.The thesis, base on researching results of image retrieval, we bring forward a method of retrievaling semantic image with the Bayes Statistical Learning Theory in hope of retrieving image with higher accuracy and efficiency by bridging the semantic gap between the visual features of the substratum of images and the meaning of images. This article aims at:(1) Setting up semantic model and finding methods in presenting and retrieving semantemes of image through studying semantic models(2) Putting forward, through investigating strategies of dividing images and retrieving feature of images, an improved algorithm in the division of images to increase efficiency of procesing images.(3) Proposing, through reserching technologies of categorizing semantic, a semantic categorizer based on Bayes Statistical Learning Theory to label images automatically.(4) Bringing forth the strtagies of designing framework and functions of the system and its databases based on reserching the prototypical system of images semantic retrieval.