Recently, much attention has been focused on new multimedia applications such as geometric information systems, multimedia medical information systems, electronic newspapers, electronic libraries, and video-on-demand services, etc. For storing and querying the multimedia data, the traditional approach is a human to attach the annotations manually then query on these additional information. But there exists several problems. In order to use multimedia data more efficient, it is very important to multimedia applications to provide the content-based retrieval which helps users to retrieve desired multimedia data for their contents.
In this thesis, we propose a method of knowledge-assisted content-based retrieval strategy. Our database can store several features to be extracted from multimedia data such as images and video clips. Also, we have designed another database for storing the domain knowledge that makes query the multimedia data easier in various ways. Finally, we have implemented the method of knowledge-assisted content-based retrieval using the UniSQL object-oriented database management system with MS-Visual C++ language on MS-Windows NT.