12noon, ECE 118
Food will be served
TITLE:
Scale Invariant Raster Image Representation Through Topological Encoding
by Warren Hunt, graduate researcher, Advanced Graphics Lab
ABSTRACT:
Raster images have proved themselves to be a pervasive and necessary
medium, however their fixed resolution and band-limited "digital" nature
impose significant limitations. Vector graphics formats achieve resolution
independence and can represent discontinuous image features, but at the
cost of real-time performance. A desire for interactive and real-time
performance, coupled with a need for scale invariant rendering, drives the
search for improved 2D and 3D dataset imaging techniques.
This work leverages a topological encoding of a dataset to attain scale
invariance from a rasterized representation. This technique, called IStar,
is a novel application of topology that captures the semantic
characteristics in a graph-like structure. This structure, along with a
reparametrized and sampled representation of the segmented dataset, is
packaged as a standard raster image which can then be substantially
downsampled and compressed. To view or render an IStar image, the encoded
image is upsampled and the topological structure is used to reconstruct
the original dataset. Unlike traditional raster approaches, this
representation can preserve sharp discontinuities at any level of
magnification, much like scalable vector graphics. Because this technique
is raster-based, it is well suited to the real-time rendering pipeline.
This technique is demonstrated, along with several extensions of the
rendering function made possible by the topological encoding.
This work was done in collaboration with Joe Kniss and Pradeep Sen.
BIO:
Warren Hunt is a New Mexico native with a diverse professional background.
Originally trained in communications, Warren has worked as a television
broadcast technician, disk jockey, and video editor. While working in the
Los Angeles Post Production industry, Warren's clients included Walt Disney
Studios and several television shows. His background also includes
designing and building contemporary hardwood furniture, and serving
champagne as old as he is, in some of New York's finest restaurants. While
completing his Masters degree in Computer Science, Warren develops and
supports visualization software at Sandia National Laboratories. His
research interests are in Scientific Visualization and High Performance
Computing. Warren is the father of two children and enjoys living close
to the Rio Grande river.
--
Pradeep Sen
Assistant Professor
Advanced Graphics Lab
Dept. of Electrical & Computer Engineering
University of New Mexico
_______________________________________________
gfx-cafe mailing list
gfx-cafe@lists.eece.unm.edu
http://lists.eece.unm.edu/mailman/listinfo/gfx-cafe
No comments:
Post a Comment