The Key Technology Research On Image Semantics Retrieval And Classification

With the rapid development of multimedia technology and computer network technology, the multimedia image data have been growth massive. And the Internet has entered an era of big data. Image retrieval based keyword is less efficient and the speed of keyword is slowly. The image tags can not also find the semantic information which is hidden in the multimedia data. In order to improve the efficiency and accuracy of image retrieval, image retrieval based content instead of keyword.This paper research on the problem of content-based image retrieval that is the semantic gap. In order to adapt for the rapid development of the multimedia market and the needs of people in the mass data environment image retrieval, this paper research on semantic content analysis and classification algorithms. The main content of this paper include the following three parts:1) This paper published a fusion algorithm based on the difference of laying. The algorithm fused the global feature to filt the irrelevance image firstly. To establish the effective image low-level feature extraction mechanism, the algorithm also can select the important information automatically. Therefore, the algorithm can achieve effective extraction and improve the efficiency of the system;2) Image semantic analysis conducted in-depth study and research on bag of words model, and improve the codebook generated algorithm by bringing in the concept of information entropy to measure the amount of 3) information on each codebook.This paper construct an optimal codebook space by merging and spliting codenote;4) Relevance feedback is a kind of human-computer interaction, relevance feedback algorithm for image retrieval algorithm to be corrected in this paper and improved particle swarm algorithm to be used for finding the optimal correction method.