and
r the
d for
tion.
ont is
ically
least
as an
s not
nsors
scale
y 337
clude
igular
drift)
sition
y the
ting a
in a
ixels.
stance
cy of
te co-
rable
cless,
X the
ut by
ASP).
ltered
nd the
| pre-
st and
s step
; with
nally,
2 into
erable
im. A
dated
were
icts in
lataset
b).
rea of
s only
crater
1. The
left a
an be
small
b the
\t first
using
| order
to reduce the computational complexity a pyramidal approach is
used. In the second step the integer disparity values are refined
to sub-pixel accuracy using a gradient based approach with an
affine adaptive window (Nefian et al. 2009). In the third step
the disparity values are transformed back to the original image
geometry and a forward intersection is calculated which results
in a 3D point cloud. In the last step this point cloud is
interpolated and map projected using a pixel size of 2 m to
derive the DTM. A shaded relief representation of the DTM is
shown in Fig. 4. As expected no matching results are available
in the shadowed areas leaving larger areas and small holes
without height values which have to be filled by other stereo
pairs and LOLA data.
s hes A %
Figure 3. Section of the left pre-rectified input image
(M139716114RE) at SRI site
Figure 4. Shaded relief of the DTM at SRI site derived from the
first stereo pair
After processing all four stereo pairs the good-quality areas are
manually selected and cropped for mosaicking. The resulting
2m mosaic is blended into an upsampled 5m LOLA DTM
(LDEM 875S 5M) in order to fill areas without matching
results and to fill holes. As can be seen in Fig. 5 the LOLA
dataset contains many outliers but this will likely change in the
future.
Figure 5. Mosaicked DTM blended into LOLA data in a
colorized shaded relief representation
4. CONCLUSIONS AND OUTLOOK
LandSAfe assists the user in selecting safe landing sites on the
lunar surface by automatically generating DTMs, detecting
craters and boulders and derived products like risk maps. In this
paper it has been shown that an automatic generation of high
resolution DTMs at the south pole of the Moon in face of
unfavorable illumination conditions is feasible. For the future
accuracy assessment processes will be included in order to
derive a confidence level for the safety of the landing site.
Additionally, the measurement of GCPs will be included in
order to obtain a better co-registration of the high resolution
DTM and the LOLA dataset.
S. REFERENCES
Acton, C. H., 1996. Ancillary data services of NASA's
Navigation and Ancillary Information Facility. Planetary and
Space Science, 44(1), pp. 65-70.
Anderson, J. A., Sides, S. C., Soltesz, D. L., Sucharski, T. L,
Becker, K. J., 2004. Modernization of the integrated software
for imagers and spectrometers. 35th Lunar and Planetary
Science Conference, March 15-19, League City, Texas, abstract
#2039.
Archinal, B. A., Rosiek, M. R., Kirk, R. L., Redding, B. L.,
2006. The Unified Lunar Control Network 2005. U.S.
Geological Survey, Open-File Report 2006-1367 - Version 1.0.
477
