‘or SPOT
ges are
ing the
it is
epipolar
of SPOT
airs of
n, 1987)
8 problem.
ata. Zhou
in which
arranged
fitting
e Zhou s
tion and
images
as a
188 not
image to
level
ipproach
or left
| master
nomial
ted by
ponding
tic or
e test
riation
by the
viously
cessing
ly the
D. So
cessing
ts for
y a few
irallax
Ig." "in
lax is
i0 the
id be
image
mented
4.
rallax
id be
the
n the
T image,
tching
r and
4, INTERSECTION AND ANALYSIS OF EXPERIMENTS
RESUILTS
It should be noted that the elevation of grid
point is calculated by using the exterior
orientation elements of the line in which the
corresponding image point is located. According
to the new data organization method, the
following preparation must be made before
intersection: the conjugat points coordinates
are those of original image coordinates system,
the exterior orientation elements of central
line and its variation rate have been
calcalated, the set of the grid points must
form essentially orthophoto cases,
Tab.3 illustrates the erperiment results of
elevation information extraction from test area
1. It can be seen that RMS in three dimentional
direction are all less than 10 m, which is
calealated in 23 check points.
6. CONCLUSION
It is feasible to utilise the SPOT stereopairs
to extract the elevation information by means
of fully digital processing manner. In the
solution of exterior orientation elements, some
techniques which can be applied to improve the
accuracy of the solution should be taken into
use. Multilevel image matching is implenented
effectively and reliably, especialy, the
modified LSM algorithms (SLSM&DPFM) are proved
to be better than tranditional LSM algorithm
in some extent. Intersection is carried out in
regular grid to produce DEM directly which
avoid the interpolation procedure and eliminate
the accuracy loss in interpolation. Consequently,
the accuracy of elevation extracted by fully
digital processing manner can reach 10 m, as
long as the the quality of image is assured.
Direction | RMS (m |
X | 9.1 |
Y 8.5
Z | 9.9 i
Table: 3 Accuracy Test for Intersection
REFERENCE
Ackermann, F., 1983. High Precision Digital
Image Correlation . Proceedings of 39th
Photogrammetric Week, Stuttgart.
Dowman, I.J., etal., 1987. Digital Processing
of SPOT Data, Proceedings on Fast Processing
of Photogrammetric Data . Interlaken ,
Switzerland, PP. 318-330.
Multi-information and
Mnlti-criterion Image Matching. Acta
Geodetica et Cartographica Sinica, No. 2,
PP. 109-114. (English version)
Lin Zongjian, 1988.
Tao Chuang, Lin Zongjian, 1992. Multi-level
Image Matching for SPOT Data. Remote Sensing
of Environment, Vol. 7, No.2, (in Chinese)
Zhou Yueqin, 1988. A New Approach to Arrange
the Approximate Epipolar Lines for SPOT
images. Acta Geodetica et Cartographica
Sinica, No.2, PP.46-61. (English version)
Zhang Zuxun, Zhou Yueqin. 1988. The Solution
of Exterior Orientation Elements for SPOT
Images. Journal of Science and Technology,
Wuhan, No.2. (in Chinese)
I
registration rectification
| size
image pairs
(pixel? control points check points control points
| =
nua. Ox gy num. OX Oy nun. Cx Cy
| 1024 |
test area 1| 741024 9-0.35 0.93 983. 0.68 2.04.
| pixel pixel pixel pixel
test area 2| 12310 5! 9 0.32 0,7 355 0.57 1.29 6 1,24 2.94
| < 1100 | pixel pixel pixel pixel | (m) (m)
Table:1 Accuracy of registration and rectification
| LSM SLSM | PDFM
window | ;
size iteration cpu-tize accuracy iteration cpu-tine accuracy | iteration ocpu-liue accuracy
(pixel) tines (second) (pixel? times (selond) (pixel) tines (second) (pixel)
8x9 5.00 1.44 2.24 3.00 0.12 1.41 | 2.48 0.93 1.85
1313 5.30 2.13 1.53 i 2.33 0.16 1.23 | 2.78 0.58 1.46
1717 5.20 4.02 1.19 | 2.58 0.23 1.01 2.00 2.05 1.11
25x25 4.40 6.85 0.77 2.57 9.39 0.94 2.40 3.31 0.74
33 x33 4.80 10.73 0.52 2.68 0.60 0.88 | 2.68 5.40 0.50
Table:2 Comparison on iteration times» cpu-time and accuracy
( NOTE: the values in the table are the average results of 119 matching points)
877
