Based On Global Histogram And Spatial Relationships Image Retrieval Technology

In recent years, with the rapidly development of image database, the number of its image in the database is surging in the traditional TBIR (text-based image retrieval) is more and more cannot satisfied with the needs of the development of the network, so the consequent researchers have proposed CBIR (content-based image retrieval) technology. CBIR is mainly used in the low-level visual features of image, such as shape, texture, color and space as a retrieval tool. In this a series of characteristics, the most widely used and most important is the extraction of color features. As compared with other traits, it to retrieve the size of the image, size and rotation are not rigid requirements, and the extraction of color features is relatively easy and convenient. In the color feature retrieval algorithm commonly used are:the reference color table method, the main colors and color histogram method.This article mainly around the color features this feature extraction in image retrieval, image retrieval system introduced in the key technology of low-level visual feature extraction, and joined the modified on the basis of this algorithm. In this paper, the main research contents and contributions are as follows.First of all, the development course of CBIR and retrieval has made a brief introduction to the key technology, including:image retrieval at present stage often used in feature extraction such as shape, texture and space. And mainly for the purpose of this article will use the color of the feature extraction of the two kinds of color space, color histogram and color quantization method done in-depth research and discussion. In addition also introduces several kinds of common to objective evaluation method of the retrieval system.Secondly, this paper focus on the global color histogram in RGB space were studied, and aimed at the present stage space histogram of color feature points of the deficiency existing in spatial relations has carried on the detailed analysis and research, and proposed an improved algorithm. After global histogram of image retrieval, the color feature point extraction, for the feature points in the image of the location distribution similarity matching relationship, to improve the defects of the feature point location is not sensitive. Later in this paper, the improved retrieval algorithm is designed for this model of content based image retrieval system, this paper expounds the structure and function design realization and the corresponding display interface, retrieval through practical test, the improved algorithm can obtain more intuitive retrieval effect.Finally, the article for CBIR (content-based image retrieval) technology algorithm research and implementation of a general summary. And for the purpose of this article the present situation of work done, and may face in the next phase retrieval technique problems and the focus of research work gives the analysis and research, put forward the development direction of new outlook for the future.