gives:
(9)
inally (9)
terms of
to get an
(10)
Iculated,
ortant is
at if we
an 10%
lude (c),
| propor-
d to the
e edges
1e height
1e MISR
ines, full
eference
in areas
J., cloud
efficients
is check
n of the
d (12) to
ed toler-
herwise,
peat the
Blunder Detection: A blunder detection function was imple-
mented to prevent low accuracy and extra sub-girding effort
caused by the appearance of blunders from image match-
ing. The reliability of the image-to-image transformation was
studied by examining the behavior of the cofactor matrix Q,,,
of the residuals. The distribution of the control points are
configured accordingly. Blunders are detected using a data
snooping method by a combined test to the standardized
residuals v; and to the variance per unit weight o, .
Reduced Number of Tie Points: As described previously,
the tie points coordinates are obtained using an image
matching technique in addition to the supplied navigation
and attitude data. However, ceratin conditions (e.g. deserts,
cloud cover) in a MISR image will not allow for successful
matching in those region. In those cases the corrections to
the supplied navigation and attitude data are modeled as
slowly varying parameters. These parameters are obtained
through the use of a Kalman filter based on the information
from the tie points previously matched.
Band-to-Band Transformation: The registration between
the new MISR image and ROI imagery has been done using
the red spectral band (Figure 3) because of its characteris-
tics relative to the image matching requirements. The imag-
ery from the other three bands will be registered to the
already registered and geolocated red band. This registra-
tion does not include image matching. Rather, the transfor-
mation between bands is defined by the interior orientation
parameters of the geometric camera model. More details on
this transformation may be presented in a subsequent
paper.
5. PROTOTYPE TESTING
Delivery of the beta version of the production software was
in March of 1996, following an extensive prototype and test-
ing phase. Landsat TM images and associated DEM have
been used to produce simulated MISR data, and navigation
and attitude data errors are included (Lewicki, 1994). Sev-
eral test cases have been made with two objectives: 1) to
represent a realistic range of perturbations and errors in the
navigation and attitude data, and 2) to represent various
cases in regards to the availability of a region suitable for
image matching. Only in the worst combination of these fac-
tors (i.e. worst possible errors in the attitude, less then 5096
of a region suitable for matching, and no information from
the previous matching available) does our algorithm not
meet geolocation accuracy requirements, which is not sur-
prising. Otherwise, testing has demonstrated that georectifi-
cation of MISR imagery which meets science accuracy
requirements is feasible in an autonomous and continuous
process.
6. ACKNOWLEDGMENTS
The authors gratefully acknowledge the efforts of the MISR
Principal Investigator, David J. Diner and the members of
the Science Data System Team: Graham W. Bothwell, Earl
181
G. Hansen, Kenneth L.Jones, and Scott A. Lewicki. Addi-
tional thanks are due to Carol J. Bruegge and Robert P.
Korechoff for their efforts in MISR camera calibration and
Nevin A. Bryant for his involvement in the creation of a glo-
bal DEM. This research is being carried out at the Jet Pro-
pulsion Laboratory, California Institute of Technology, under
contract with the National Aeronautics and Space Adminis-
tration.
7. REFERENCES
[1] Ackermann, F., Digital Image Correlation: Perfor-
J
Photogrammetric Record, vol. 11(64), 1984.
[2] Diner, D. J., Bruegge, C. J., Martonchick, J. V., Both-
well, G. W., Danielson, E. D., Floyd, E. L., Ford, V. G.,
Hovland, L. E., Jones, K. L., and White, M. L., A Mul-
tiangle Imaging SpectroRadiometer for Terrestrial
Remote Sensing from the Earth Observing System,
International Journal of Imaging Systems and Tech-
nology, vol. 3, 1991.
[3] Herrick, S,. Astrodynamics, Van Nostrand Reinhold,
New York, 1971.
[4] Forstner, W., On the Geometric Precision of Digital
Correlation, ISPRS Int. Arch. of Photogrammetry, vol.
XXIV, Commission Ill, Helsinki, 1980.
[5] Lewicki, S. A., Smyth, M. M., Jovanovic, V. M., and
Hansen, E. G., A Simulation of EOS MISB Data and
Geometric Processing for the Prototyping of the
MISR Ground Data System, IGARSS Proceedings,
vol (3), 1994.
[6] Mikhail, E. M., Observations and Least Square,
Harper & Row, New York, 1976.
[7] Paderes, F. C., Mikhail, E. M., and Fagerman, J. A.,
ery, ASPRS Proceedings, vol. (3), 1989.
[8] Snyder, J. P., Map Projection - A Working Manual,
United States Geological Survey Professional Paper
1395, U. S. Government Printing Office, Washington,
1987.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996
