Based Close To The Degree Theory, Image Database Retrieval Technology

Content based image retrieval has been a hot focus in the field of the research for image retrieval. Essence of CBIR is acquainting and extracting the base feature of the image, and then detecting the image extracted whose methods come in a variety of sizes and styles. In order to describe the similar and near degree of the sets of similar feature between two images, this paper discusses the relationship of the measures of similar standard and presents its theorem, property and relevant certificates. In the framework of close-degree theory, this article discusses and contracts the computational methods of two close-degrees; improves the traditional computational methods of close-degree; demonstrates its practical application of the measurement of similarity.To describe histogram into digital form, this paper suggests a definition and description of methods for color characters granular computing based on regional histogram. The methods not only to maintain the Inherent characteristics of the original global histogram, but also to a certain extent, reflect the characteristics of spatial distribution of the image color. Feature set is also used to describe the shape characteristics of the image, and puts forward a close-degree calculation method based on feature vector which comes from Hu moments describing the geometry. On the basis of various granular set in the images using similarity measure theory leading to granular concept close-degree, we calculate these particles set and arrange obtained values so as to achieve the purpose of retrieval. The reason why many retrieval system or methodology is not conducive to real-time calculation and rapid analysis is often using a more complex feature vector and calculation methods to improve the retrieval precision. In this paper, the proposed feature vector search method can greatly simplify the feature vector and calculation methods, has more suitable form for real-time computing which not only can improve the efficiency and stability of image retrieval, and also has relatively good expression of the experimental data in the aspect of accuracy.