Research On Content-Based Image Retrieval In World Wide Web

Nowadays, internet has become an important resource of information. As a result, effective methods of information retrieval are needed extraordinarily. As an important part of multimedia information retrieval, Content-Based Image Retrieval(CBIR)has been one of most active research areas in the past few years. Now, many breakthroughs was achieved in this field, and many business and research systems are developed.Content-Based Image Retrieval is a technique for retrieving images on the basis of image features such as colors, textures and shapes, etc. The main contents of this paper include color analysis, image segmentation, image database construction and image retrieval and so on. In our CBIR system, we searching image from internet by Search Engines, and extracting features from raw image. People can query image by those features such as color,shape and so on.Color is one of the most important image features. The paper presented a new colorspace named HS'B, in which features of human vision are more considered than in other colorspace. In this paper, all of that has relation to color is based upon this colorspace, including Dominant Color Features extraction and image object color analysis.The paper gives a new arithmetic of Dominant Color Features extraction. In HS'B colorspace, we can get the algorithm by clustering pixels of one image. We pick up several peak values from the algorithm as dominant colors of the image. The result of experiment on a large number of images shows that the arithmetic is efficient and effective.[mage segmentation is a critical steps of image analysis. Object features of one image could be extracted only after the image was segmented.Two principal method of image segmentation are based on regions and edges lie in the image. The paper discuss both of the two method and present two arithmetic based on each of them. Furthermore, a post treatment method of segmentation is given in the paper. Accordingly, we do many experiments on scenic picture and receive satisfied result.In the paper, the class of image database and character of image Database Management System(DBMS) are discussed in detail.Accordingly,a relation database is constructed. The paper classified image data into three kinds, namely primary features, vision features and semantic features. These three features describe one image from different levels.In the end of the paper, an image manipulation and analysis demo system is developed, and a CBIR system is designed.The system platform for developing these systems is Windows 2000, and the development environment is Visual C++ 6. 0. The paper perform two image retrieval experiments. One is based on color features. The other is based on both of color and shape features. The results of the experiments are satisfying.