Study Of Content-Based Image Retrieval Methods In Multimedia Data

Multimedia database is a new database technology to adapt to Multimedia data storage and search etc. In this article, related concept and technology of MMDB are discussed, and the multimedia data type in common uses, the new problems, system structure, data model, and implement methods of MMDBMS are also analyzed. Content-based Retrieval is one of the most important methods of query and research in Multimedia data, and it is also the research hotspot. At the same time, the concept, feature, system structure, process, search methods of CBIR, and classify of query interface, form of query specification are discussed. I also analyze feature extraction of Multimedia data, search index methods, search methods etc. On the base of analyzing research direction, application status and main technology of the current CBIR system, I design and realized a prototype of small CBIR system. In this system, because the use of combined methods of text-based and content-based retrieval, the shortcoming of complex and long-time calculation is got over, the speed and veracity are improved. CBIR is a gradually precision process. There is a feature adjustment and rematch process in it. Technology is one of the most important methods in improving query. According to relative feedback theory, I put forward a single, straight feedback approach, namely by set similitude degree between every image in query result and target image, the query quality is improved.