] Canada -
pace.gc.ca
an.gc.ca
eveloped for
{ven if three
om 1:20,000
^re obtained.
l. The DEM
en compared
0% (LE90),
“land covers
re achieved.
results were
m with -2 m
) were better
obtained for
de modèles
1 par 10 m).
. PAs stéréo
1,9 m et 4,7
nterne de la
nages multi-
| obtenu des
te la surface
du sol, on a
ptenues. De
des résultats
t LE90 de 5
copique, les
ation stéréo
-1’s Optical
ydular Opto-
nann et al,
ission and
, 2002) and
py (HRS)
pabilities of
erate digital
s multi-date
sical model
developed at Canada Centre for Remote Sensing (CCRS) for
medium-resolution sensors in the visible and infra-red (MODIS,
MERIS, Landsat, SPOT, ASTER, etc.) as well as in the
microwave (SIR-C, JERS, ERS-1, RADARSAT, ENVISAT)
(Toutin, 1995), was adapted these last years for high resolution
data, such as SPOT-5 HRG across-track data (Toutin, 2004). A
preliminary version has been recently developed for SPOT-5
HRS in-track stereo-data and is used in this study.
2. STUDY SITE AND DATA SET
2.1 Study Site
The study site is an area north of Québec City, Québec, Canada
(47° N, 71° 30° W). This study is an urban, rural and forested
environment and has a hilly topography in the south with a
mean slope of 7°, and mountainous topography in the north with
a mean slope of 10° and maximum slopes of 30°. The elevation
ranges from 0 m at the St-Lawrence River to 1000-m in the
Canadian Shield. Québec City is in the south-east part.
Figure 1. SPOT-5 HRS fore image, acquired north of Québec
City, Canada (120 km by 60 km; 10 m by 5 m pixel
spacing). The yellow box represents the across-track
stereo-pair (60 km by 60 km) and the green box the
Lidar (5 km by 13 km).
SPOT-5 © 2003 CNES and Courtesy SPOT-IMAGE
2.2 Data Set
The +22° in-track stereo-images (120 km by 60 km; 10 m by 5
m pixel spacing; base-to-height ratio, B/H, of 0.85) were
acquired September 18, 2003 as a courtesy of SPOT-Image,
France with 5% of clouds and their shadows (Figure 1). The
SPOT-5 images are raw level-1A data, orbit oriented, with
detector equalization only. Ephemeris and attitude data are
available in the metadata as well as general information related
to the sensors and satellite.
In addition, SPOT-5 HRG across-track stereo-pair (Figure 1
yellow box; 60 km by 60 km; 5 m by 5 m pixel spacing; B/H of
0.77) was acquired on May 5 and 25, 2003 with viewing angles
of +23° and -19°, respectively. The May 5 image displays snow
in the forests (upper part) and frozen lakes (lower left and
417
International Archives of the Photogrammetry, Remote Sensin g and Spatial Information Sciences, Vol XXXV, Part B1. Istanbul 2004
centre), for almost 50% of the image, but not the May 25
image. These differences in snow/ice generated large
radiometric differences in SPOT stereo-images. However,
these differences provide an opportunity to test DEM
generation method and address potential problems in difficult
conditions instead of working in a perfect environment.
To evaluate the accuracy of the stereo-extracted DEMs,
accurate spot elevation data was obtained from a LIDAR survey
conducted by GPR Consultants (www.lasermap.com) on
September 6", 2001 (Figure 1 green box). The Optech ALTM-
1020 system is comprised of a high frequency optical laser
coupled with a Global Positioning System and an Inertial
Navigation System. The ground point density is about 300,000
3-D points per minute and the accuracy is 0.30 m in planimetry
and 0.15 m in elevation. Since it was impossible to cover the
full SPOT stereo-pair (60 km by 120 km), ten swaths covering
an area of 5 km by 13 km (Fig. 1) and representative of the full
study site were acquired. The results of the LIDAR survey are
then an irregular-spacing grid (around 3 m), due also to no echo
return in some conditions such as buildings with black roofs,
roads and lakes. Since the objectives of this research study
were to evaluate the stereo DEMs, the LIDAR elevation data
was not interpolated into a regular spacing grid so as to avoid
the propagation of interpolation error into the checked elevation
and evaluation.
3. EXPERIMENT
Since the processing steps of DEM generation using either in-
track or across-track stereo images are well known, the six
processing steps are summarized in Figure 2 (Toutin, 1995):
1. Acquisition and pre-processing of the remote sensing data
(images and metadata) to determine an approximate value
for each parameter of 3D physical model for the two
images;
2. Collection of stereo GCPs with their 3D cartographic
coordinates and two-dimensional (2D) image coordinates.
GCPs covered the total surface with points at the lowest
and highest elevation to avoid extrapolations, both in
planimetry and elevation. Ninety-eight and thirty-three
GCPs were acquired for in- and across-track stereo-pairs,
respectively from 1:20,000 topographic maps (2-3 m
accuracy in the three axes). The image pointing accuracy
was less than one pixel.
3. Computation of the stereo models, initialized with the
approximate parameter values and refined by an iterative
least-squares bundle adjustment (coplanarity equations)
with the GCPs (Step 2) and orbital constraints. Both
equations of colinearity and coplanarity are used as
observation equations and weighted as a function of input
errors. Theoretically three accurate GCPs are enough to
compute the stereo model, but more GCPs were acquired
either to have an overestimation in the adjustment and to
reduce the impact of errors or to perform accuracy tests
with independent check points (ICPs).
4. Extraction of elevation parallaxes using multi-scale mean
normalized cross-correlation method with computation of
the maximum of the correlation coefficient. This method
gave good results and was commonly used with satellite
VIR images (Gülch, 1991);
5. Computation of XYZ cartographic coordinates from
elevation parallaxes (Step 4) using the previously-
