Applications to Lunar Remote Sensing - 3D Computer Vision: Efficient Methods and Applications - page 309

A more recent HRSC-based global topographic map of Mars is described by
Gwinner et al. ( 2010 ); it has an average lateral resolution ('grid spacing') of 93 . 1m
and exhibits an average absolute difference with respect to the MOLA topographic
map of 29.2 m and a corresponding root-mean-square deviation of 49 . 9m.
8.1.1.4 Photoclinometry and Shape from Shading
While the lateral resolution of a topographic map generated by stereophotogram-
metric techniques is generally significantly lower than the pixel resolution of the
images from which it has been produced, the technique of photoclinometry de-
scribed in Sect. 3.2.2.1 can be used to generate cross-sectional profiles of plane-
tary surfaces at a resolution comparable to that of the available images. It has been
pointed out in Sect. 3.2.2.1 that photoclinometric methods yield a fairly accurate
representation of depth differences on small scales but tend to produce systematic
depth errors on scales much larger than the pixel resolution. Early work by Wilhelms
( 1964 ) describes the photoclinometric determination of the statistical distribution of
small-scale slopes on the lunar surface, using telescopic photographic lunar images.
Photoclinometry is used by Mouginis-Mark and Wilson ( 1979 ) to measure crater
depths on Mercury, relying on Mariner 10 images. Howard et al. ( 1982 ) apply a
photoclinometry technique to Viking orbiter images of the northern polar region of
Mars, revealing elevation variations of a few hundred metres on horizontal scales of
about 10 km.
More recent work by Fenton and Herkenhoff ( 2000 ) describes the determination
of topographic data for the 'layered deposits' situated in the northern polar region of
Mars. They use three-dimensional points determined by stereophotogrammetry as
an absolute reference for three-dimensional profiles obtained by photoclinometry,
where a constant surface albedo is assumed and an approximate correction of the
influence of the atmosphere is performed. The results are compared with MOLA
data. The photoclinometric measurements reveal thicknesses of individual layers of
the order of some tens of metres. Beyer and McEwen ( 2002 ) use a photoclinome-
try technique to examine the suitability of selected regions on the Martian surface as
landing sites for rover missions by determining surface gradients on spatial scales of
several metres. Herkenhoff et al. ( 2002 ) estimate the roughness of the 'residual ice
cap' in the northern polar region of Mars, i.e. the ice persisting during the Martian
summer, in order to examine its suitability as a landing site. They determine the sta-
tistical distribution of surface slope angles inferred by photoclinometry. Lohse et al.
( 2006 ) apply their 'multi-image shape from shading method' (cf. Sect. 5.4 ) to con-
struct topographic maps of the lunar surface based on Clementine images. A similar
approach is used by Gaskell et al. ( 2007 ) for a three-dimensional reconstruction of
the asteroid Eros, relying on Near-Earth Asteroid Rendezvous (NEAR) spacecraft
image. Grieger et al. ( 2008 ) propose a photometric stereo approach for constructing
a DEM of the region around the lunar south pole based on several partially overlap-
ping images acquired by the SMART-1 AMIE camera under different illumination
conditions.