tanbul 2004
M
Ruiz, who
so go to the
ion at the
ving us the
t Müller of
1e German
a software
"Accuracy
ges using a
ternational
ig, Vol. 33.
deoTeX-: a
ternational
1g, Vol. 29,
ation study
ace project
ternational
ig, Vol. 29,
Zobler D.,
ternational
ynsemantic
ery — the
ch, Heipke
| 3-87907-
p. 86-102.
t of Digital
he — 43th
orithm for
computing
ndbook, S-
2
Lu
f SPOT-S
SPOT-5 HRS ALONG-TRACK STEREO DATA EVALUATION
OVER MONTMIRAIL AND MELBOURNE TEST SITES
R. Nandakumar , Amitabh, Medha Alurkar, B. Gopala Krishna, P.K. Srivastava
Space Applications Centre, Indian Space Research Organisation, Ahmedabad — 380 015, India —
(nand, amitabh, medha, bgk, pradeep)@ipdpg.gov.in
Commission I, WG 1/2
KEYWORDS: SPOTS, stereo imaging, digital elevation model, DEM evaluation, IRS-1C/1D stereo software
ABSTRACT:
As part of CNES-ISPRS HRS Study Team investigation, SPOT-5 HRS along-track stereo data over Montmirail (France) and
Melbourne (Australia) test sites were evaluated using the modified IRS-1C/1D stereo photogrammetric software Saphire and PCI
Geomatica Orthoengine (version 9) software. Both the stereo pairs were processed using the four corners and one center pixel
planimetric coordinates for each individual image as control points.
The linearly interpolated satellite state vectors were used in the space resection module of the modified Saphire software to compute
the updated attitude angles based on the control point coordinates, separately for each image. Conjugate points were automatically
identified using interest points based hierarchical image matching. Using the space intersection module, (planimetric coordinates
and) heights were computed at each conjugate point. These were interpolated to generate a uniform DEM grid of required spacing.
For Montmirail, the DEM generated was compared with the two reference DEMs provided in Lambert I] projection with NTF datum
in Clarke 1866 ellipsoid, which are accurate to about 1 m. For Melbourne, the DEM could not be generated using the PCI Geomatica
possibly due to large cloud patches. It is observed that without using any control points, the modified Saphire software could achieve
an accuracy of 20 m at 89% (10 m at 75% and 5 m at 58%) of the DEM points in the case of Montmirail. The results could be
improved if manual interaction for interactively increasing the density of conjugate points and identifying break-lines are introduced
in Saphire. Availability of a near-nadir image would also help reduce the occlusions that occur in high slope areas.
1. INTRODUCTION
This paper discusses the study carried out at Space Applications
Centre (SAC) on evaluating the along-track stereo capability of
the SPOTS HRS sensor using its data sets over two test sites,
Montmirail, France and Melbourne, Australia; over which
precise control points and/ or precise pre-existing digital
elevation models (DEM) are available. The study uses the in-
house developed stereo photogrammetry software Saphire and a
commercially available software PCI Geomatica for DEM
generation; and another commercially available software
ERDAS Imagine for DEM interpolation and evaluation.
1.1 SPOT-5 Mission & HRS Instrument: SPOTS, the latest in
the series of French optical remote sensing satellites, was
launched on May 4, 2002. Refer (Vadon, 2003). It has two
identical and independent HGR (high geometrical resolution)
instruments, having an improved resolution of 5 m and 2.5 m in
panchromatic mode and 10 m in multispectral mode. It has a
vegetation instrument VGT identical to that of SPOT 4 for
global vegetation monitoring. It has a new high-resolution
stereoscopic (HRS) instrument for the production of digital
terrain models.
HRS instrument consists of two fixed cameras, which are
inclined fore-ward and aft-ward by + 20° along the ground
track. It has a base-to-height ratio of 0.8. The HRS instrument
operates in the panchromatic band with a swath of 120 km. The
pixel size on ground is 10 m x 10 m. However, pixels are
sampled every 10 m perpendicular to the ground track, but
every 5 m in the along track (epipolar direction) direction so as
to increase the altimetric accuracy.
Localization precision of an HRS image is about 25 m rms, the
specification indicated being 50 m rms. Localization precision
of the digital terrain model produced by HRS is about 10 m
(after check using ground control points). Altimetric accuracy
of digital terrain model produced from a pair of images is 5 m
rms (Vadon, 2003).
1.2 CNES-ISPRS HRS Study Team: ISPRS and CNES had
selected 9 test areas for the HRS Scientific Assessment
Program, which was announced during the ISPRS Denver
Symposium in November 2002 (http://www. isprs.org/). The test
areas were selected from among the 40 proposed sites, based on
the availability of cloud free HRS stereo pairs and Ground
Reference Data in the form of control points and/or DEM of
sufficient accuracy offered by the Principal Investigators (PI).
Refer (Baudoin et al, 2003) for more details on the HRS
Scientific Assessment Program initiative.
2. DATA SETS USED
2.1 HRS Stereo Data Sets:
Data Product Type : SPOT system SCENE level 1A
Format : DIMAP
Geometric Processing Level : RAW
Raster data Format * GEOTIFF
: 12000 & 12000
: Panchromatic
: 5050-261/1 02/08/14 10:42:46 IS
&5050-261/1 02/08/14 10:44:18 2S
:5 379-426 03/02/19 00:23:28 1 S
& 5 379-426 03/02/19 00:25:01 2 S
No. of scanlines & pixels
Spectral Band
Montmirail Scene Ids
Melbourne Sene Ids
