Tasks:
- Data management - The uniform processing
and archiving of data derived from a variety of remote sensing
sensors will be imperative to ensure repeatability in analyses and
in building confidence in the decision systems. Towards this
goal it will be necessary to develop standard protocols for
processing data for use across the RSTC. Also, to address the
issue of handling the potential terabytes of data being
generated from this research, two needs assessments will be
initiated. One will be to assess data compression algorithms
applicable to remote sensing data, and the other will be to
determine the viability of available intelligent data-mining
algorithms for use in remote sensing database management.
- Decision systems - Much of the
cross-cutting work with agriculture, forestry and wildlife, and
transportation will be in this area. Research in decision systems
is a three-step process:
- Develop an understanding of how targets interact with the
electromagnetic spectrum (phenomenological research with focus
area).
- Construct algorithms or models that describe this
interaction and permit classification.
- Develop a suite of automated tools to expedite consistent
image analysis and enhance consumer confidence.
Core research projects in this area will utilize agriculture and
forestry data and will be focused in the first two steps of the
process.
Geospatial
visualization - This task will focus on
developing a suite of visualization tools to permit researchers
and end users to attain an understanding of the information
content of remote sensing images. This involves visualization
toolkits optimized for lower-end desktop workstations, large
lab-style graphic workstations, and computerized virtual
environments (COVE) for 3D immersive visualizations. Often, these
toolkits will be developed for use over the Internet. This will
permit remote users the opportunity to interact meaningfully with
large databases of images (e.g., MS statewide forest inventory).
- Capacity building - At
present, most research in the focus areas has been limited to
passive sensor technologies. Microwave remote sensing is
an active technology that brings new information about a target
to aid in classification. This effort will focus on establishing
strategic partnerships with industry that will lead to the
acquisition of the appropriate hardware to establish a microwave
remote sensing laboratory at MSU.
|
|
Deliverables (divided into the three Task categories):
Data
management
- A standard protocol applicable to the three focus areas for
processing of multispectral and hyperspectral images.
- A report making recommendations on the use of data
compression and data mining algorithms for management of the
RSTC’s remote sensing data archives.
Decision systems
- Demonstration of the effectiveness of extant sensor models
for use in the virtual mode of the remote sensing end-to-end
information channel.
- An analysis technique for detecting reflectance and
absorption bands in hyperspectral data.
- A mixing model and subpixel classification algorithm that
describes soil, weed, and crop interactions.
- A methodology for determining site specific soil sampling
points via remote sensing.
Geospatial visualization
- A capability to read any digital elevation map (DEM)
dataset and fly through it interactively in the immersive (COVE)
environment.
- A suite of analysis tools for interactively examining
various scenarios of classified images via the Internet.
- Initial development of the capabilities to reconstruct 3-D
visualizations of tree stand structure from multi-return imaging
LIDAR data.
Capacity building
- Specifications and strategic partnerships with companies to
realize a microwave remote sensing laboratory at MSU.
|
PREV
PAGE
