Effective Algorithm Using Similarity Based Technique for Image and Video Databases

Navidgator, an intelligent browsing-tool for image and video data is developed. The main problem with the handling of multimedia databases is the navigation through and the search within the content of a database. The major problem arises from the difference between the possible textual description (annotation) of the database content and its visual appearance. Overcoming the so called semantic gap has been in the focus of research for some time. In the present study, a new system for similarity-based browsing of multimedia databases is presented. The system aims at decreasing the semantic gap by using a tree structure, built up on balanced hierarchical clustering. Using this approach, operators are provided with an intuitive and easy-to-use browsing tool. In the present study an attempt is made is not only on the description of the database organization and retrieval structure, but also how the proposed techniques might be integrated into a single system. In this paper an attempt is made for the direct use of a balanced tree structure for navigating through the database of keyframes, paired with an easy-to-use interface, offering a coarse to fine similarity-based view of the grouped database content.