# control R.M.S.E.
points X [m] Y [m] Z [m]
28 18.34 11.53 14.94
Table 2 : Model accuracy for the Sor Rondane images
The low overall accuracy is mainly caused by the diffi-
culty by which reference and control points could be
identified in the images. The control points on the im-
ages were taken on the contact surface between rock
outcrops and ice/snow surfaces, which was done very
accurately. However, the contact line between rock and
snow on the topographic maps at scale 1:50,000 is so
strongly generalised, that geographic coordinates could
not be identified properly.
At the moment we are in the possession of a SPOT
stereocouple of the northwestern part of Corsica.
These images will be used for further analysis of the
planimetric accuracy, once enough reliable ground
control is available.
4.3. Matching Accuracy
The matching accuracy is the percentage of correctly
matched points compared to the total amount of
matches performed. Matched points are evaluated on
the basis of the correlation coefficient and the heuristic
77007
Lj
y
ZZ
BD
AR
a,
2
2%
IL
ES
<Z
iX
ZZ
7
ed
doc
e
y
LJ
ZZ
GIE
2
Of
=
SESS,
2X5
CS
A
7
ee
525
SM
ZZ
/
2%
2602
2
P
T
ou
a]
V
»
SONA %
Rte EDS S E
S a
eZ S AL —
A 2% PASSI C 2100
509 22 ee LUCI $256 2000
LAH ee
13 LY NONE —— 1900
OV LILI LISE ROZ
DI S W IR — 1800
ILLS Y Es
IISG A So 9 1700
euo S Io I 2 ZH d
RO LS 7 Al ^-^ 1600
8250 PS WS — 1500
J - 1400
REL Z
RIZ 1300
… 1200
— 1100
PEL
ZZ
FLD
E
knowledge (see 4). For the Antarctic scenes, matching
accuracy did not exceed 60%, both because of the steep
relief and therefore large shadows and a higher detec-
tor saturation for the east-looking image. Also, only a
few matched points were accepted (on the base of the
aforementioned criteria) on the ice and snow surface.
In order to improve the matching accuracy over high
reflectance zones, selective image enhancement tech-
niques (high pass filters and contrast stretching) on the
glacierised surface were applied, resulting in a far
much better performance (accuracy of 70-80%).
However, on saturated snow patches matching still re-
mains impossible.
Fig.5 displays a generalised DEM of the Menipa region
(rock outcrop in the Ser Rondane surrounded by sev-
eral glaciers) obtained from stereo-matching from
SPOT. Heights range from 1100 m (flat glacier surface
in front) to 2100 m. Preliminary matching tests could
not distinguish corresponding points on the glacier
surface, due to the low contrast. However, selective
enhancement improved results a lot.
476
5. CONCLUSIONS
A method is presented in order to derive surface to-
pography from stereoscopic SPOT images in polar re-
gions. Despite the low precision by which reference
points could be identified in the images, the overall
accuracy still allows the production of topographic
maps at scales between 1:100,000 and 1:50,000. More ac-
curate ground control should allow for higher preci-
sion and therefore larger scale mapping. Matching ac-
curacy and computation efficiency were improved by
adding constraints to the matching algorithm concern-
ing geometric characteristics and terrain texture and
implying selective image enhancement techniques.
Results obtained this far prove that topographic map-
ping from space is undeniable the solution for an effi-
cient production of medium to large scale topographic
maps in remote (polar) areas.
6. ACKNOWLEDGEMENTS
This research is part of the “Belgian Scientific
Research Programme on Remote Sensing by Satellite -
phase two” (Services of the Prime Minister - Science
Policy Office).
7. REFERENCES
Chevrel, M., M. Courtois and G. Weill, 1981. The SPOT
Satellite Remote Sensing Mission.
Photogrammetric Engineering and Remote
Sensing, 47(8): 1163-1171.
CNES and SPOT IMAGE, 1988. SPOT’s User
Handbook, Vol.1: Reference Manual. CNES and
SPOT IMAGE (Eds.).
Hottier, P and M. Albattah, 1990. SPOT et la
Photogrammétrie Amnalitique Solution de
Quelques Problémes. Bulletin SFPT, 117(1): 10-23.
Leclerc, A., 1989. Le Systéme CARTOSPOT. CISM
Journal ACSGC, 43(2): 111-120.
Li, M., 1991. Hierarchical Multipoint Matching.
Photogrammetric Engineering and Remote
Sensing, 57(8): 1039-1047.
Pattyn, F., 1991. The Geocoding of SPOT Imagery in
Raster-Based Geographical Information Systems. In:
Proceedings of the second European Conference on
Geographical Information Systems EGIS'91,
Brussels, Belgium, April 2-5, 1991, pp. 841-46.
Rosenholm, D., 1985. Digital Matching of Simulated
SPOT Images. Fotogrammetriska Meddelanden,
2(50): 1-76.
Salamonowicz, PH., 1986. Satellite Orientation and
Position for Geometric Correction of Scanner
Imagery Photogrammetric Engineering and
Remote Sensing, 52(4): 491-499.
Ungar, S.G., C.J. Merry, R. Irish, H.L. McKim and M.S.
Miller, 1988. Extraction of Topography from Side-
Looking Satellite Systems - A Case Study with SPOT
Simulation Data. Remote Sensing of Environment,
26: 51-73.
Westin, T., 1990. Precision Rectification of SPOT
Imagery. Photogrammetric Engineering and
Remote Sensing, 56(2): 247-253.
Wong, R.Y. and E.L. Hall, 1978. Sequential Hierarchical
Scene Matching. IEEE Transactions on Computers,
C27(4): 359-366.
