Segmentation and Visualization of Volume Maps
Powei Feng
Abstract
Volume data is a simple and often-used representation for exchanging and processing
data in various scientific domains, such as medicine and molecular biology. The
segmentation of volume data is an essential part of data interpretation. Researchers
have extensively studied the problem of segmentation focusing on efficient algorithms
for segmenting and rendering volumes. Our contribution is two-fold. First, we propose
a tri-linear classification method that can implemented on the GPU to reduce artifacts
and jaggedness along the material boundaries that appear when rendering segmented
volumes. Our representation provides sub-voxel accuracy for representing segmented
materials. Second, we demonstrate our interactive painting-based segmentation tool,
which can be used to rapidly produce an intuitive segmentation. We compare our tool
against known results and show that we can generate similar segmentations using a
simple and intuitive control scheme.