Indexes for User Access to Large Video Databases

by:

Lawrence A. Rowe, John S. Boreczky, and Charles A. Eads
Computer Science Division - EECS
University of California Berkeley
Berkeley, CA 94720
(rowe@cs.berkeley.edu)

 

Abstract

Video-on-Demand systems have received a good deal of attention recently. Few studies, however, have addressed the problem of locating a video of interest in a large video database. This paper describes the design and implementation of a metadata database and query interface that attempts to solve this information retrieval problem.

Sample queries were collected by interviewing a variety of users. These queries guided the design of indexes that can be used to answer the types of queries users want to ask. Three types of indexes were identified: 1) bibliographic (e.g., title, abstract, subject keywords, genre, director, cast, etc.), 2) structural (e.g., segments, scenes, shots, transitions, sound effects, etc.), and 3) content (e.g., keyframe sequences, scripts, objects and actors in scenes, etc.).

A mixed-mode query interface was built that allows a user to select a set of videos and/or frame sequences and incrementally modify the answer set. The interface includes relational, hierarchical browsing, and keyword search operations.