Peer Reviewed Articles
891 Delineating Forest Canopy Species in the Northeastern United
States Using Multi-Temporal TM Imagery
John G. Mickelson, Jr., Daniel L. Civco, and John A. Silander, Jr.
Abstract
Download
Full Article
We generated a detailed forest type map of the dominant
conopy species within northwestern Connecticut using multiseasonal
Landsat Thematic Mapper (TM) data which were
ground referenced with the Global Positioning System (GPS).
The map was designed as a calibration layer for a spatially
explicit forest dynamics model we have developed, called
SORTIE, and will allow us to test the model's effectiveness in
predicting landscape level patterns. The precisely located
field data were used to derive the forest class signatures used
in the classification. Combining the six reflective bands each
from spring, summer, and fall Landsat TM images to create
an 18-band composite allowed for genus level forest classification
precision. We delineated a total of 33 forest classes:
20 dominant types with 13 additional sub-classes representing
differing understory composition. Accuracy assessment
using the Gopal-Woodcock fuzzy set process returned an
overall forest class accuracy of 78.9 percent at the procedure's
Acceptable level.
905 Spectral Shape Classificationof Landsat Thematic Mapper Imagery
Mark J. Carlotto
Abstract
Download
Full Article
A multispectral classifier based on an alternative spectral
representation is described, and its performance over a full
Landsat Thematic Mapper (TM) scene is evaluated. Spectral
classes are represented by their spectral shape - a vector of
binary features that describes the relative values between
spectral bands. An algorithm for segmenting or clustering TM
data based on this representation is described. After classes
have been assigned to a subset of spectral shapes within
training areas, the remaining spectral shapes are classified
according to their Hamming distance to those that have already
been classified. The performance of the spectral shape
classifier is compared to a maximum-likelihood classifier
over five sites that are fairly representative of the full Landsat
scene considered. Although the performance of the two
classifiers is not significantly different within a site, the performance
of the spectral shape classifier is significantly better
than the maximum-likelihood classifier across sites.
Analysis of results suggest that the spectral shape classifier
is relatively insensitive to seasonal changes between wetland
and upland areas in the scene and is not affected by thin
clouds aver one of the sites. A full-scene spectral shape classifier
is then described which combines spectral signature
files that associate classes with spectral shapes derived over
the five sites into a single file that is used to classify the full
scene. The classification accuracy of the full-scene spectral
shape classifier is shown to be superior to that of a stratified
maximum-likelihood classifier.
915 Responses of Spectral Indices to Variations in
Vegetation Cover and Soil Background
Stella W. Todd and Roger M. Hoffer
Abstract
Download
Full Article
The primary objective of this study was to evaluate the effects
of variations in soil texture and moisture upon the
green vegetation index (GVI) and the normalized difference
vegetation index (NDVI) for targets with specific vegetation
cover amounts and varying soil backgrounds. The second objective
was to understand the difference in information provided
between NDVI and GVI relative to estimating vegetation
cover. The third objective was to investigate the information
contained within the wetness/brightness plane in relation to
sail background characteristics and variations in percent
canopy cover. Brightness and wetness were estimated using
the Tasseled Cap brightness index (BI) and wetness index
(WI).A simple two-component model of soil and green vegetation reflectance was used to simulate the effects of three
soil texture types (sand, silt, and clay) and two soil moisture
classes on greenness, brightness, and wetness values.
The results indicated that, for the same vegetation percent
cover class, targets with more moist soil backgrounds
displayed higher NDVI values than did targets with more dry
soil backgrounds. In contrast, GVI values were much less influenced
by soil background variation. WI values increased as
green vegetation cover increased far all soil backgrounds.
The largest increase was for dry soil backgrounds. BI values
either increased or decreased as green vegetation cover increased,
depending on soil background brightness. BI and WI
provided complimentary spectral information.
923 A Technique for 3D Building Reconstruction
Taejung Kim and Jan-Peter Muller
Abstract
Download
Full Article
An approach to tackle the problem of three-dimensional (3D)
building reconstruction in urban imagery is presented. For 3D
building reconstruction, there is a need to combine 2D (such
as grouping) and 3D analysis (such as stereo matching). A "good" strategy for the combination is essential for success.
A simple but robust combination strategy is proposed. Combination
is carried out only after a 2D building detection
technique and a 3D height extraction technique are applied
completely independently. The 2D building detection technique
does not use any information generated from the
height extraction technique, nor vice versa. Moreover, any
assumptions or conditions derived in the course of 2D building
detection or height extraction are not used for combination.
3D building reconstruction is done by interpolating
heights into the area covered by 2D building boundaries using
the 3D height information. In this way results from the 2D
building detection technique and 3D height extraction technique
can be meaningful by themselves. This also can make
the process of 3D building reconstruction simple and applicable
to a wide range of images. This approach is tested with
airborne images, and the results show that 3D building reconstruction
can be achieved successfully.
| Top | Home |