Abstract: This talk covers several aspects of compressing geometric data sets.
First, the need for such simplification & compression algorithms is motivated
using examples from various applications (i.e. motion planing, field navigation
and GIS expert systems). Then a brief survey on existing terrain simplification
methods is given. 

The main part of the talk covers a new algorithm for terrain simplification.
The new method is based on known Digital Image Processing compression methods
(e.g. DCT, wavelets compression) which were specially adjusted to fit
Digital Elevation Models. DEM-images are terrains or elevation maps represented
as gray scale images. Based on the special nature of terrain-images a unique
pre-compression process is designed. A set of benchmark experiments comparing
the new suggested terrain simplification algorithm (named ICTS*) with other
well known methods will be presented. The main conclusion form this experiment
is that the ICTS algorithm leads to significantly better compression results,
faster runtime, and most importantly, for a given file size it represents
the original terrain better than other terrain simplification methods.

At the last part of the talk a set of generalizations for the ICTS algorithm
will be presented, in particular: A frame-work for compressing a natural &
urban elevation maps.  An algorithm for compressing Z-cam geometric data
(see for example http://www.3dvsystems.com), and a natural open problem of
generalizing the ICTS to 3D-4D volumetric geometric datasets such as (MRI &
FMRI bio scans).

*Patent pending IL182367