A Prototype Content-Based Retrieval System that Uses Virtual Images to Save Space

Previous research has demonstrated that space can be saved in a MultiMedia DataBase Management System (MMDBMS) by storing some of the data items virtually, meaning they are stored as sequences of editing operations. The existing approaches for performing Content-Based Retrieval (CBR) in an MMDBMS, however, typically assume that the data items are stored as large, binary objects. The result is that an MMDBMS cannot use the existing approaches without losing the space savings gained by storing the data items virtually. In our demo, we present a prototype CBR system for virtual images that avoids this problem by using the semantic information in the editing operations during retrieval. 1