International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B4, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
one of which is designed for the visible bands and the second
for the ultraviolet (UV) wavelengths. At the base of the UV
optics is a prism that redirects the light to a common CCD
mounted under the visible optics. Seven narrow-band
interference filters bonded over the detector array enable the
acquisition of the color images.
The camera is designed to operate in two modes: monochrome
and color. In the monochrome mode (nominally the 643nm
band) the WAC acquires framelets that have 1024 samples and
14 lines. In color mode, the WAC acquires framelets for all
seven bands, however due to limitations in the readout rate of
the CCD array, only the center 704 samples are read out for
each 14 line visible band. For the UV framelets, the center 512
samples are read out of the UV portion of the detector array.
During the read out, the 512 samples and 16 lines are summed
in 4 x 4 pixel boxes resulting in a 128 x 4 pixel framelet, which
increases the signal to noise ratio for the UV bands.
Due to this configuration, in monochrome mode the WAC has
a ~90° FOV and in color mode a ~60° FOV. The nadir
pointing pixels have a pixel scale of 75 meters for the visible
bands from an altitude of 50 km, while UV bands have a
pixel scale of 384 meters from the same altitude due to the
summing. All framelets imaged simultaneously are stored as
one frame. The WAC repeatedly acquires frames at a rate
such that each of the narrow framelets overlaps providing
continuous coverage for each color band. Typical WAC
observations contain 36 to 1,800 evenly spaced frames.
Figure 2. LROC WAC before spacecraft integration
2. PRE-LAUNCH CALIBRATION
Prior to launch, the geometric distortion of both camera systems
were characterized post-assembly at Malin Space Science
Systems (MSSS). Each camera was mounted on an Ultradex
rotary stage that provided “azimuthal” control in one degree
steps with an accuracy of one arc second. The two NACs
imaged a bar pattern, while the WAC, with a second rotary
stage to control the elevation, scanned a collimated spot of
selectable size over a range of azimuth and elevations. By
recording the angles of each observation and registering
them to corresponding pixels in the array, a geometric model
for each camera was constructed. Each model was defined
by a focal length, boresight, and radial distortion
coefficients. After calibration, the NAC was securely mounted
through the center of the spacecraft’s optical bench using three
bolts connected near the focal plane. The WAC was mounted on
the front of the optical bench (+Z, Observation side) on a small
pedestal to keep the wide FOV clear of all obstructions and to
offset the radiator from the optical bench.
3. IN-FLIGHT CALIBRATION
Accurate placement of NAC and WAC images in a cartographic
framework requires precise knowledge of the camera
orientations relative to the spacecraft coordinates and distortion
of the optics. Using a subset of the vast image dataset collected
by LROC, improved pointing and distortion parameters were
derived. This section describes the methodology and
improvements to the geometric calibration.
3.1 NAC Pointing Correction
Due to the twin camera configuration of the LROC NAC, there
are three components to the geometric calibration:
1. Absolute twist offset: A rotation about each camera's
boresight, determined relative to the spacecraft frame.
2. Absolute offset: The offset of a given map-projected
pixel from its true coordinates. This is a rotation about
the two axes perpendicular to the camera's boresight
(cross-track and down-track).
3. Relative offset: The temperature-dependent
component of the offset between the left (NAC-L) and
right (NAC-R) cameras. Like the absolute offset, this
is a rotation about the axes perpendicular to the
camera's boresight.
The products of these components were combined to provide a
precise (seamless NAC-L to NAC-R registration) and accurate
(within ~20 m to surface coordinates) camera model for
projecting the immense NAC dataset available to the science
and engineering community.
3.1.1 Absolute Twist Correction: The two NAC cameras
are nominally mounted such that the sensors are parallel to each
other and perpendicular to the flight direction. Comparison of
several images of Apollo sites revealed that NAC-L and NAC-R
images, once projected, were rotated relative to each other 0.36°
to 0.40° (Figure 3). As an independent check, twist values were
derived relative to projected Descent Stage of the Apollo Lunar
Module (LM) and retroreflector locations (in the same image) at
the Apollo sites: the twist offset was 0.37°.
The absolute value of each camera rotation relative to the
spacecraft reference frame was determined by creating a control
network from over 3,800 polar NAC images [Lee et al., 2012].
The orientation of these images varied, making it possible to
derive a twist adjustment. The average derived absolute
rotational offsets were -0.24° for the NAC-L and +0.13° for the
NAC-R, for a relative rotation of 0.37°.
4 4 fs
x10 x10 Offset (p)
-25 -25
: *
-3 3
35 35
4 4
45 45 =
-
-5 5
55 5.5
-6 6
EN y
Figure 3. Down-track (left) and cross-track (right) offsets
between a NAC-L and NAC-R with roughly the same footprint,
plotted against line and sample, before twist correction. Y-axis
units are 10* lines.
430