Multiresolution ray tracing for point-based geometry
Loading...
Date
2007
Authors
Zakaria, Mohamed Nordin
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The primary concern in this thesis is with the incorporation of multiresolutionbased
optimization into ray tracing algorithms specially tailored for point-based
geometry. In recent years, increasing demand for model complexity has led to an
increasing use of 3D scanning technologies capable of digitizing a complex
physical model into a dense, massive and unstructured point cloud. Despite the
dense amount of information contained in this data set, work in human
perception study has shown that not all of it will be perceptible by a human
viewer. Hence multiresolution technique offers an opportunity to reduce the
computational workload involved in ray tracing such data set. In this respect, the
first contribution in this thesis is the adaptation and enhancement of a Bounding
Volume Hierarchy data structure in order to allow for faster ray tracing. The
resulting data structure incorporates an efficient Level-of-Detail control and
backface-culling optimization, and uses a mixture of bounding spheres and
boxes to enable faster ray tracing. The second contribution in this thesis is an
approach for incorporating multiresolution-based optimization into a point-based
geometry ray tracer that is already optimized by use of a cost-optimized kd-tree.
The resulting data structure incorporates an efficient Level of Detail control and a
new variant of ray-surface intersection scheme that improves the ray tracing
performance. Both the image quality and the ray tracing performance obtained
point to the effectiveness of the multiresolution techniques introduced in this
thesis.
Description
PhD
Keywords
Computer Science , Multiresolution , ray tracing , point-based geometry