 |
ARCHIVED - Data Not Current
For Information Purposes Only
Goddard
Space Flight Center
1992 Phase II
Parallel
Processing for Aerial/Satellite Image Understanding
LNK
Corporated, Inc.
Riverdale, MD
|
INNOVATION
Developed parallel algorithms on a
SIMD architecture parallel computer for performing real-time aerial
image understanding tasks
|
Classification over the Los Angeles Area
|
ACCOMPLISHMENTS
- Used the wavelet transform as
a powerful spatio-temporal descriptor of the multispectral imagery
- Derived a neural network classifier
for performing automatic land use classification with up to 94%
accuracy
- Ported time-critical pieces
of the algorithm to the CNAPS® massively parallel
array processor
- Developed systems for automated
registration and multi/hyper spectral image compression
|
COMMERCIALIZATION
- Developed and successfully marketed
the parallel image processing library LNK_ImageLib®
- LNK_ImageLib®
has led to increased market exposure and a related contract for
LNK Corporation
- More than 50% of the revenue
from LNK_ImageLib® comes from exports
|
GOVERNMENT/SCIENCE
APPLICATIONS
- Component of the satellite image
processing package being integrated by NASA Goddard for distribution
to remote data centers (RDCs)
- Used in a parallel Navy program
for classifying and locating targets in reconnaissance imagery
- A version of the algorithm is
used in an ongoing Phase II effort with the U.S. Army for localizing
and navigating unmanned vehicles
|
| For more
information about this firm, please send e-mail to company
representative
Return
to NASA SBIR Success Listings
|
Curator:
SBIR Support 4/15/05
|
