MEE I ^ Are
AT 1 = EA PL r2 4
S 3 A PE A c
À. , STA EE
J one CE f | ST | Ve
a hi M g
= ty = rx Qo e Rats
/ A ME ß adis "n CR
cn Y ATO MENTON
AS LEE : DCUM
/ KB y T ES
y € I Y ? ^ CH, {
= ^ oo
CAEN = EE EE ERES EI
à E S t i 2 x ev t 7
1 A. CN d. N
+ A i x a, 3 s.l. "A
le. fe I XX X =
N "RA 3 f J TONS E
dor 3 = DUR, V a le tas |
I + fib: "m. 1
So MACC NL. CE = "
I vd vig h "UA. p =
E } oY; E ZA €
J FATS,
APE Ee IN CLAS
/ : RS 2 : A S
7 = ON in
ge = = = 7. S (d 3 Rte À
- À 7 Pace ES SE
+ a r
= = ; c
3 ; ~~ En d 3 25 i
A or wa
: 2, 7 E , ac x ^ d
; TEMA. s LA s
; E MAREC At ;
| S de: A Ia
z x 7
ZI 78 2 \
=
"Fame + +
+ + + +
Figure 9 Triangulation mesh for DTM interpolation of map sheet 1225
7. CONCLUSIONS
A matching algorithm for SPOT images was presented
that uses the SPOT geometry to impose constraints that
reduce the search space from 2-D to 1-D. The algorithm
severely reduces the problems caused by radiometric
differences can simultaneously use any number of images
(more than two) and determines in on step pixel and
object coordinates. The use of gradient images magnitude
instead of grey level images improves the results. A
conformal transformation is suggested. However, the use
of only shifts also leads to good results, slightly inferior
to those of the conformal transformation.
Problematic cases like multiple solutions and occlusions
are reduced and the computation times decreases due to
the 1-D search. An intelligent blunder detection scheme
is proposed that uses criteria derived solely from the
statistics of the results.
The accuracy of the matching is excellent (RMS less than
10 m for 36,000 check points). The accuracy of the
interpolated DTM depends clearly on the density of
measurement points and for sufficient density it can also
be in the 10 m range.
924
8. ACKNOWLEDGEMENTS
The authors express their gratitude to the Bundesamt für
Landestopographie, Bern, Switzerland for providing the
SPOT images and DTM data and appreciate their
cooperation in the accuracy analysis of the results.
9, REFERENCES
Baltsavias, E. P., 1992, Multiphoto Geometrically Con-
strained Matching. Ph. D. Dissertation, Mitteilungen Nr.
49, Institute of Geodesy and Photogrammetry, ETH Zu-
rich, 221 p. /
Baltsavias, E. P., Stallmann, D., 1992. Metric information
extraction from SPOT images and the role of polynomial
mapping functions. In: Proc. of 17th ISPRS Congress,
Commission IV, 2 - 14 August, Washington D. C., USA.
Kratky, V., 19892. On-line aspects of stereophotogrammet-
ric processing of SPOT images. PERS, Vol. 55, No. 3, pp.
311 - 316.
Kratky, V., 1989b. Rigorous photogrammetric processing
of SPOT images at CCM Canada. ISPRS Journal of Pho-
togrammetry and Remote Sensing, Vol. 44, pp. 53 - 71.
