Multiresolution modeling using binary space partitioning trees

Abstract

Space partitioning techniques are a useful means of organizing geometric models into data structures. Such data structures provide easy and efficient access to a wide range of computer graphics and visualization applications like real-time rendering of large data bases, collision detection, point classification, etc. Binary Space Partitioning (BSP) trees are one of the most successful space partitioning techniques, since they allow both object modeling and classification in one single structure. Also, due to the fact that complexity of 3D models is increasing far more rapidly than the performance of graphics system, there is an increasing need for multiresolution modeling techniques. This paper presents a novel method that extends BSP trees to provide such a representation. The models we present have the advantages of both BSP trees and multiresolution representations. Nodes near the root of the BSP tree store coarser versions of the geometry, while leaf nodes provide finer details of the representation. The goal of this work is to build a single tree that provides a high number (nearly a continuous range) of representations of an object at different resolutions, with minimum redundancy. This model is especially well suited for been used within Internet 3D graphics applications as it provides for efficient progressive transmission and fast (hardware independent) rendering of tridimensional scenes.