Peer Reviewed Articles
109 SPOT and Landsat Stereo Fusion for Data Extraction over Mountainous
Areas
Thierry Toutin
Abstract
In thematic and cartographic applications, planimetric features
are extracted from multi-sensor images such as SPOT
and Landsat in order to take advantage of their complimentarity
features. When no precise elevation data are available
to ortho-rectify these images, e.g., in mountainous areas,
stereo digital photogmrnmetric workstations (stereo DPWS) are
now available for the interactive stereo fusion and plotting of
multi-sensor stereo pairs.
This paper presents a method and the quantitative results on the extraction of planimetric and altimetric features from a stereo pair generated with mixed sensor images (SPOT-P and Landsat TM) using the stereo DPWS, the DVP available at the Canada Centre for Remote Sensing. Results from this mixed sensor stereo pair, which has a base-to-height ratio of 0.49 and was taken over a mountainous area of the Rocky Mountains (Canada), show a planimetric accuracy of 11 to 12 m for well identified cartographic features, and an altimetric accuracy of 37 m for the extracted elevation data.
115 Mosaicking of Orthorectified Aerial Images
Yehuda Afek and Ariel Brand
Abstract
Aerial photographs are widely used in surveying, geographic
information systems (GIS), and other applications. Analysis of
a large area requires the creation of an image mosaic, which
is composed of several aerial photographs. In an ideal situation,
, a perfect mosaic can be generated using a series of
rigid transformations on the source images. In practice, geometric
distortions and radiometric differences interfere with
the mosaicking process.
In this paper a complete algorithm to mosaic images taken at different times and conditions with geometric distortions and radiometric differences is presented. The algorithm, which works without any human intervention, integrates global feature matching algorithms into the process of selecting a seam line. The algorithm may be applied to mosaic any set of images for which an appropriate matching algorithm exists.
The creation of an image mosaic is accomplished using local tmnsformations along a computed seam line and a rigid transformation elsewhere. An automatic stereo matching algorithm, originally developed for surface height measurement, is used to detect matching pairs of tie points across frame boundaries. These tie points are used to compute the seam line for the mosaic, and to compute geometric and radiometric correcting transformations around this seam line.
127 Estimating the Accuracy of Coarse Scale Classification Using High
Scale Information
Christiane Klöditz, Angelien van Boxtel, Elisabetta Carfagna, and
Willem van Deursen
Abstract
A method is proposed to estimate the classification accuracy
of low-resolution images by using high-resolution Images in
place of ground truth information. For that reason, low-resolution
pixels have been simulated by degrading high-resolution
images using different multi-resolution techniques. We
carried out a statistical analysis, resulting in two models that
combine the information from the aggregated high-resolution
Landsat TM Normalized Difference Vegetation Index (NDVI)
data in order to predict NOAA NDVI. The method is illustrated
by a case study on mapping surface roughness of different
landscape classes in order to determine the amount of
deposition of atmospheric pollutants.
135 Experiments in the Identification of Terrain Features Using a PC-Bsed
Parallel Computer
Demetrius-Kleanthis D. Rokos and Mark P. Armstrong
Abstract
This paper describes the process through which an existing
serial algorithm that classifies terrain features in a digital elevation
model (DEM) is recast into a version that executes in
a parallel computer environment. The transformation process
is guided by a formal analysis of dependency relationships
that exist among the different steps in the algorithm.
The resulting program is tested using a small parallel computer
system. The results demonstrate that run times are
reduced, and that the processors are used efficiently. The
general approach to the development and implementation of
parallel algorithms that is presented in this paper is extensible
to a wide range of geogmphical computing problems.
143 A Change Detection Experiment Using Vegetation Indices
John G. Lyon, Ding Yuan, Ross S. Lunetta, and Chris D. Elvidge
Abstract
Vegetation indices (VI) have long been used in remote sensing
for monitoring temporal changes associated with vegetation.
In this study, seven vegetation indices were compared
for their value in vegetation and land-cover change detection
in part of the State of Chiapas, Mexico. VI values were developed
from three different dates of Landsat Multispectral
Scanner (MSS) data. The study suggested that (1) if normalization
techniques were used. then all seven vegetation indices
could be grouped into three categories according to their
computational procedures; (2) vegetation indices in different
categories had significantly different statistical characteristics,
and only NDVI showed normal distribution histograms;
and (3), of the three vegetation index groups, the NDVI group
was least affected by topographic factors in this study. Comparisons
of these techniques found that the NDVI difference
technique demonstrated the best vegetation change detection
us judged by laboratory and field results.
| Top | Home |