HARD
SOIL
(Pure)
SOIL
WATER
THE TOGRAPHY OF ASTEROID IDA: A COMPARISON BETWEEN
PHOTOGRAMMETRIC AND TWO-DIMENSIONAL PHOTOCLINOMETRIC IMAGE ANALYSIS
B. Giesel, J. Oberst! , R. Kirk2, and W. Zeitler!
1 DLR, Institute of Planetary Exploration, Berlin, Germany
2 ys. Geological Survey, Flagstaff, USA
Commission III; Working Group 2
KEY WORDS: DTM, Space, Asteroids, Surface, Modeling
ABSTRACT
We derived high resolution Digital Terrain Models from stereo images of the asteroid Ida that were obtained by the Galileo spacecraft
during the flyby in August 1993 and compared these results with terrain models derived from two-dimensional photoclinometry. The
comparison shows that there are striking discrepancies between the results from the two models depending on the spatial scale lenght
of surface features. While photogrammetry resolves large- and medium-scale features, photoclinometry successfully resolves small-
A TN ER REE vs
Scrubs
1996
scale features. Ideally, both methods should be combined to optimize terrain modeling.
1. INTRODUCTION
Photogrammetry and photoclinometry (commonly termed
"shape-from-shading") are complimentary methods to infer
information on surface topography from image data.
Traditionally, photogrammetry has been used to analyze aerial
stereo imagery for many decades. More recently,
photogrammetric methods also have been applied to images
acquired by spacecraft orbiting the earth, such as the SPOT
satellite and the SPACE SHUTTLE, as well as by deep space
missions. With the availability of digital image data and digital
processing techniques, our means to recover topographic data
from imagery have improved significantly. In contrast, the
"shape-from-shading" method was not fully developed until
digital and radiometrically precise image data became available.
The fact that photoclinometry can obtain topography using
single images has made this method especially popular in
planetary sciences, as space missions have data storage and
transmission capacity too limited for multiple imaging, i.e.,
stereo coverage. In addition, it is often the properties of
planetary surfaces with limited variations in albedo and the lack
of atmospheres that make shape-from-shading analyses feasible.
The basic procedure in photoclinometry is to use observed pixel
brightness to derive local slopes, from which topography can be
generated by integration. There are two different approaches to
photoclinometry: The one-dimensional model returns the
topography along a linear profile for which, however, the
topographic strike must be known in advance. The method was
originally developed to determine the topography of lunar mare
ridges the slopes of which are too gentle for the traditional
shadow-lenght method to be employed (van Diggelen, 1951). It
has since been used to analyze images from Mars (McEwen,
1985; Herkenhoff and Murray, 1990; Goldspiel et al., 1993) and
the outer planet satellites (Squyres, 1981; Schenk, 1989; 1990).
The two-dimensional approach to photoclinometry recovers the
topography from the entire image, but requires a number of a
priori knowledge of the surface properties. The technique was
suggested for use in computer vision (Horn, 1975), and has also
been used in planetary science (Kirk, 1987), but its popularity
suffers from the complexity of the approach and the
computational efforts to be taken.
In summer 1993, the Galileo spacecraft obtained images of the
asteroid Ida (Belton et al., 1994). There is quite some interest in
the topography of this planetary body, because of the
implication for its geology (Sullivan et al., 1996). In this paper,
we therefore present high-resolution terrain models of Ida
determined by methods of photogrammetry and compare these
with models derived from two-dimensional photoclinometry.
Some of the latter were previously reported (Sullivan et al.,
1996) and some are new. The results are discussed in terms of
the limitations of both methods, and it is shown how a
combined analysis could improve the results significantly.
2. DATA BASE
The Galileo spacecraft is equipped with the SSI framing CCD
camera (800x800, f=1500.467 mm). During the Ida flyby, 57
images were obtained from distances between 236,000 km and
2,500 km with resolutions ranging from 2.4 km/pixel to 25
m/pixel (Belton et al., 1994), acquired in different spectral
bands. However, the data base we used in this study comprises
36 images taken from ranges closer than 60,000 km acquired
within a time period of 81 min. During this sequence Ida
performed a rotation of 105°, which was essential for viewing
of Ida from different perspectives and thus for having good
stereo coverage.
Navigation data are available in an inertial frame with an
accuracy of 2 km for spacecraft position and 2 mrad for camera
pointing. To convert these to the Ida-fixed system we used the
rotational elements of Ida that were derived in an earlier
photogrammetric analysis of Galileo SSI images (Davies et al.,
1996).
3. METHODS
3.1 Photogrammetry
The photogrammetric analysis of the image data was carried out
in three different stages:
3.1.1 Adjustment of the image orientation parameters: The
adjustment of the image orientation parameters was performed
in the body-fixed frame of Ida (cf. Ohlhof et al., 1996). First,
we identified 95 image points in the 36 images resulting in 883
image point measurements. Then, the image point coordinates
and the observed camera navigation data were subjected to a
least squares adjustment procedure, assuming that the errors of
the observations are distributed at random. One ground control
245
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
TE TT ET
