Date Awarded

1996

Document Type

Dissertation

Degree Name

Doctor of Philosophy (Ph.D.)

Department

Computer Science

Abstract

Due to sampling processes volumetric data is inherently discrete and most often knowledge of the underlying continuous model is not available. Surface rendering techniques attempt to reconstruct the continuous model, using isosurfaces, from the discrete data. Therefore, it natural to ask how accurate the reconstructed isosurfaces are with respect to the underlying continuous model. A reconstructed isosurface may look impressive when rendered ("photorealism"), but how well does it reflect reality ("physical realism")?;The users of volume visualization packages must be aware of the short-comings of the algorithms used to produce the images so that they may properly interpret, and interact with, what they see. However, very little work has been done to quantify the accuracy of volumetric data reconstructions. Most analysis to date has been qualitative. Qualitative analysis uses simple visual inspection to determine whether characteristics, known to exist in the real world object, are present in the rendered image. Our research suggests metrics and methods for quantifying the "physical realism" of reconstructed isosurfaces.;Physical realism is a many faceted notion. In fact, a different metric could be defined for each physical property one wishes to consider. We have defined four metrics--Global Surface Area Preservation (GSAP), Volume Preservation (VP), Point Distance Preservation (PDP), and Isovalue Preservation (IVP). We present experimental results for each of these metrics and discuss their validity with respect to those results.;We also present the Reconstruction Quantification (sub)System (RQS). RQS provides a flexible framework for measuring physical realism. This system can be embedded in existing visualization systems with little modification of the system itself. Two types of analysis can be performed; reconstruction analysis and algorithm analysis. Reconstruction analysis allows users to determine the accuracy of individual surface reconstructions. Algorithm analysis, on the other hand, allows developers of visualization systems to determine the efficacy of the visualization system based on several reconstructions.

DOI

https://dx.doi.org/doi:10.21220/s2-k33s-h070

Rights

© The Author

Share

COinS