46th Lunar and Planetary Science Conference (2015)
J. B. Garvin1, K. S. Edgett2, M. R. Kennedy2, M. E. Minitti3, M.J. McBride2, G.M. Krezoski2, R. Dotson5, R. A.
Yingst3, J. Gomez-Elvira4, P.G. Conrad1; 1NASA Goddard (NASA GSFC, Greenbelt, MD 20771;
[email protected]), 2MSSS, San Diego, CA, 3Planetary Sciences Institute, Tucson, AZ, 4Centro de
Astrobiología (CSIC-INTA), Carretera de Ajalvir, km. 4, 28850 Torrejón de Ardoz, Madrid, Spain, 5Fireball at
GSFC, Greenbelt, MD.
Introduction: The Mars Science Laboratory
(MSL) [1,2] science payload includes the Mars Hand
Lens Imager (MAHLI), which has been operated in a
variety of configurations since early in the mission to
provide stereo-overlap imaging. MAHLI observations
have been used to support mission science goals
associated with rock texture and other geologic
properties as assessed in 2D images and stereo
anaglyphs [2]. Here we focus on quantitative 3D stereo
data products, in particular derived micro-relief
Quantitative Relief Model (QRM) results that permit
analysis of millimeter-scale texture. QRM’s derived
from MAHLI stereo images have been analyzed since
early in the mission (Fig. 1). In this report we
showcase a preliminary MAHLI stereo experiment
designed to assess the quality of a derived QRM and
the “noise floor” of the system by examining a pair of
Sol 808 images of the REMS UV Sensor [3] box on
the rover upper deck.
Background: 3D modeling of rock and outcrop
surfaces is commonly applied in paleontology and
archeology for non-destructive sample analysis and
artifact documentation; 3D reconstruction of bedding
features, stratigraphic and facies relations over
distance; fracture analysis; and as part of the
documentation of sample extraction sites for
reproducible science as referenced to geographic and
stratigraphic position. Micro-relief models also inform
regarding rock weathering and potential as
cryptoendolithic habitat or paleo-habitat. For Mars
surface science, not only can quantitative surface relief
models (QRMs) be applied for similar purposes (e.g.,
documentation of sample extraction sites, stratigraphic
and facies relations within and between outcrops, onEarth laboratory analysis of the 3D shape of an object),
application to habitability assessment in terms of submillimeter-scale rock surface texture (i.e., present
environment) and preservation potential for fossilized
microbial communities (i.e., past environments).
MAHLI Stereo Imaging: The focusable color
MAHLI camera can acquire sub-mm spatial resolution
images of martian geologic materials from its location
on the turret on the Curiosity rover arm. We have
examined an initial MAHLI stereo pair acquired on Sol
88 (Fig. 1) that features a discrete clast approximately
5 cm in diameter. This example reflects the potential
for measuring textures at ~0.1mm vertical scales for
geologic materials of scientific importance. To date,
MAHLI stereo pairs have been acquired for a variety
of surfaces starting with the first contact science on Sol
46, and continuing until present. Further QRM’s are
planned after rigorous analysis of the vertical error
budget intrinsic to the MAHLI stereo imaging
configurations. Our focus here is on a first-order error
analysis of QRM performance for a human-made
object on the rover deck as a validation experiment.
Initial characterization/validation experiment:
On Sol 808 a preliminary MAHLI stereo imaging
experiment was conducted to evaluate the quality of
derived QRM’s for potential further utilization during
the MSL Mission. The experiment was designed to
allow for MAHLI stereo-overlap-imaging of the
REMS UV Sensor box which is installed on the upper
rover deck, and fully described by a CAD file provided
by the REMS PI (from EADS-Astrium CRISA) [3].
Figure 1: Sol 88 MAHLI micro-relief model featuring a discrete
clast with a few cm’s of total relief. MAHLI image pixel scale is 66
µm and the frame was acquired from ~ 41 cm stand-off distance
under near-nadir viewing geometry. The derived QRM has 1 mm
grid cells and a vertical precision of ~ 0.2 mm.
As illustrated in Fig. 2, the REMS UVS box [3]
was successfully imaged from a stand-off distance of
16.8 cm under non-ideal solar illumination conditions
(i.e., with sun glints) on Sol 808 using an arm
movement to achieve a large degree of stereo overlap.
The pixel scale of the two overlapping MAHLI frames
is 66 µm. The REMS UV Sensor box is 58 mm x 68
mm and approximately 19 mm in relief above the
mounting plate on the rover deck. There are 6 circular
UV-sensing photodiodes, each 8.5 mm in diameter
arrayed on the upper plate of the REMS box [3]. Given
46th Lunar and Planetary Science Conference (2015)
the availability of the CAD model, we have been able
to generate a highly precise “QRM” of the feature for
use in absolute comparison to the QRM derived from
the MAHLI stereo pair (Fig. 2).
Figure 2: Color-scaled QRM from MAHLI stereo images
acquired on Sol 808 from ~16.8 cm above that showcases the
~ 19 mm of total relief from the deck to the top of the REMS
UVS Box (showin in Red). Black areas are those in shadow
in the stereo pair. Color scale bar to Left.
MAHLI QRM analysis: Our emphasis has been
on evaluating the “noise” floor of the computed QRM
in comparison with the 3D CAD model of the REMS
UV sensor. This initial analysis showcases the
potential vertical precision limits of the MAHLI stereo
imaging for micro-relief, and can be used as a guide
for scientists interested in using this approach for
quantitative texture analysis in the future.
The difference between the REMS UV sensor
“QRM” (from the CAD model) and the preliminary
MAHLI QRM was computed after careful registration
of the two models. Figure 3 illustrates the differences
and suggests that the “vertical noise floor” of the
MAHLI QRM is ~0.1 mm on a spatial grid scale of 0.3
mm. This indicates that textures with relative relief of
0.2-0.3 mm can be detected effectively for certain
geologic targets at mm spatial scales (for the 1st time).
Figure 3: Statistical analysis of “vertical noise” levels in the
computed MAHLI QRM. At the top is a color contoured map
of vertical errors cast as standard deviation of local relief at
0.3mm grid scales, with the color-coded MAHLI QRM shown
below. Typical values are ~0.1 mm.
Differential Analysis: It is instructive to compare
the QRM’s from the REMS UVS CAD model and the
MAHLI stereo pair after sub-grid scale registration to
examine the distribution of errors under the
illumination and geometric conditions of the Sol 808
experiment. Figure 4 depicts the distribution of local
topographic differences between the two QRM’s for
the upper plate of the REMS UVS box. For this
smooth upper plate, the measured variation is
~0.1 mm, while larger variations can be seen in the UV
photodiode “pits” mostly due to shadowed illumination
and image saturation. More than 80% of the statistical
differences are < 0.2 mm on a grid scale of 0.3 mm
(Fig. 4). This suggests that martian rock sample
textures at such scales can be reliably measured,
especially if illumination conditions are optimized.
Figure 4: Sol 808 MAHLI QRM differential relief analysis. At top
are perspective views of the REMS UVS box from the CAD model
next to the MAHLI QRM. A profile through the MAHLI QRM is
shown in the middle panel relative to the CAD model. A histogram of
local topographic variations between the CAD model and the
MAHLI QRM is in the lower panel. See text for details.
Conclusion and Future Work: Even under nonideal imaging conditions, the preliminary MAHLI
QRM for Sol 808 illustrates vertical precision levels
better than 0.2 mm. Quantitative analysis of such
QRM’s is a step beyond routine, qualitative anaglyph
evaluation, and offers the possibility of new boundary
conditions for micro-relief analysis of rock textures.
References: [1] Grotzinger, J. P. et al. (2012) Space
Sci. Rev. 170, 5–56. [2] Edgett, K.S., et al. (2012),
Space Sci. Rev., doi:10.1007/s11214-012-9910-4. [3]
Gómez-Elvira, J. et al. (2012) Space Sci. Rev. 170,
583–640, doi:10.1007/s11214-012-9921-1. [JBG is
grateful to Drs. J. Grunsfeld and C. Scolese for support].