Content-based Image Retrieval

With wide applications of multimedia database such as the Digital Library, the management and query of image database pose an obstacle to such applications. Accordingly, content-based image retrieval has been the focus of the research since the 1990s.As known to all, content-based image retrieval is mainly stimulated under the combination of pattern recognition and multimedia database.This paper firstly discusses the extraction of color and texture feature. Then this paper provides an image data model which is used to manage and query the images. Lastly, this paper designs a content-based image retrieval system in order to evaluate the effectiveness of previous algorithms provided in this paper.The emphasis of this paper:1. discussed many algorithms about color extraction, and provided the extraction of main color.2. discussed many algorithms about texture extraction, and firstly applied the texture spectrum into image retrieval. Experiments indicate texture spectrum outperforms gray co-occurrence.3. provided an image data model based on relation model and object Oriented database, and applied it into image retrieval .