skip to main content

Title: A hyperspectral image analysis workbench for environmental science applications

A significant challenge to the information sciences is to provide more powerful and accessible means to exploit the enormous wealth of data available from high-resolution imaging spectrometry, or hyperspectral'' imagery, for analysis, for mapping purposes, and for input to environmental modeling applications. As an initial response to this challenge, Argonne's Advanced Computer Applications Center has developed a workstation-based prototype software workbench which employs Al techniques and other advanced approaches to deduce surface characteristics and extract features from the hyperspectral images. Among its current capabilities, the prototype system can classify pixels by abstract surface type. The classification process employs neural network analysis of inputs which include pixel spectra and a variety of processed image metrics, including image texture spectra'' derived from fractal signatures computed for subimage tiles at each wavelength.
Authors:
; ; ;
Publication Date:
OSTI Identifier:
6987920
Report Number(s):
ANL/CP-77460; CONF-930141--1
ON: DE93000603
DOE Contract Number:
W-31109-ENG-38
Resource Type:
Conference
Resource Relation:
Conference: 9. international conference on interactive information and processing, Anaheim, CA (United States), 17-22 Jan 1993
Research Org:
Argonne National Lab., IL (United States)
Sponsoring Org:
DOE; USDOE, Washington, DC (United States)
Country of Publication:
United States
Language:
English
Subject:
99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE; 54 ENVIRONMENTAL SCIENCES; SITE CHARACTERIZATION; IMAGE PROCESSING; SURFACES; COMPUTERIZED SIMULATION; FRACTALS; LAND USE; PROCESSING; SIMULATION 990200* -- Mathematics & Computers; 540210 -- Environment, Terrestrial-- Basic Studies-- (1990-)