Querying large virtual models for interactive walkthrough

Abstract

This thesis addresses database issues for real-time walkthrough of large Virtual Environments. The main purpose of the research is to design novel techniques to facilitate storing, manipulating, and retrieving of dataon secondary storage, and to provide optimization support for real-time rendering.We propose novel algorithms, such as caching, prefetching, and Complement Searching, based on spatial index and design the REVIEWwalkthrough system, which implements these spatial techniques. We also propose a novel disk data structure called Hierarchical Degree-of-Visibility (HDoV) tree, and several algorithms based on the tree structure. We also test the performance of the memory version of the HDoV-tree structure. These visibility techniques are implemented in the VISUAL system. We conduct extensive performance experiments on these two systems.