Research On Content-Based Image Retrieval Technology

Most of the prevailing image retrieval systems concern the key word of the image description, easily neglecting another important element of image- the content of image. Content-Based Image Retrieval (CBIR) is a kind of technique for retrieving images based on automatically extracting visual features such as color, texture, and shape etc. It has been an active research area in recent years. In this paper, the key technology on color-based and shape-based of CBIR is researched. The main points in this thesis are listed below:(1) Based on the technology of color-based image retrieval, first, we improve the dominant color histogram using histogram and normal distribution method; then, according to the characteristics of the image layout, we improve the histogram which based on sub regions and primary colors, using a fast image retrieval algorithm. The two algorithms can not only better express the image color composition, and fully consider the spatial information, but also improves the retrieval speed.(2) Based on the technology of shape-based image retrieval, using contour- tracing method to fix the main target area in order to extract shape feature. Based on this, we adopt a multi-feature integrated retrieval method, which fully considering both color and shape feature, greatly enhancing the recall ratio and precision ratio of image retrieval.(3) In order to testify the validity of the algorithm, we design a CBIR prototype system. We introduce the system structure, modules and function, the realization of the users interface and all algorithms experiments' analysis.The paper's innovation:(1) By anglicizing the experimental results of several improved algorithm, we have verified the effectiveness of the algorithm and greatly improved the image retrieval performance on recall ratio and precision ratio.(2) The CBIR prototype system in this paper adopts C/S three-layer structure. The improved algorithm in the middle web layer facilitates researchers to study or call the algorithms; the user-oriented layer that provides a user-friendly interface, simplifys the user's operation. The system has good adaptability and expansibility.