Declustering Objects for Visualization

In this paper we propose a new declustering method which is particularly suitable for image and cartographic databases used for visualization. Our declustering method is based on algebraic techniques using vectors. The algorithm which computes the disk assignment requires O(Kj log K) time where K is the number of parallel disks in the system. The resulting disk assignment maximizes the area that can be visualized without accessing any disk more than once. The method is easy to implement and works for any number of parallel disks. Our mathematical analysis show that for common visualization queries our declustering method performs within seven percent from optimal for a wide range of practical multiple disk configurations.