Clementine HiRes Data Processing
Statement of Work
Malin Space Science Systems (MSSS) will create overlapping and/or
strip mosaics of the Clementine HIRES images. HIRES images were
systematically acquired poleward of 50° latitude, at each pole,
and on each orbit. The HIRES instrument had a field of view of
0.4° by 0.3° ; the swath width near the poles was about 10 km
in the apoapsis hemisphere and about 5 km in the periapsis hemisphere.
Thus, overlapping coverage (using data from both months) only exists
for latitudes above about 82° . At a resolution of 1/1024
degrees/pixel (29.6 meters/pixel) the two polar mosaics will fit on a
single CD-ROM. The remainder of the systematic HIRES data form
non-overlapping strips. MSSS will produce calibrated, registered
image strips, at the average resolution of the strip, using a
sinusoidal map projection. The total volume of these strip mosaics
will be about 18 GB, or 28 CD-ROMs.
Construction of mosaiced products from Clementine data will be
performed systematically, and MSSS will initially concentrate its
efforts on automated mosaicing techniques. The U. S. Geological
Survey is using a hybrid approach (both automated and manual) to
construction of the UVVIS mosaics and MSSS will interact closely with
and benefit from this effort, which already is receiving NASA support.
For each area to be mapped, the processing steps required are:
- Locate all images covering the area to be mosaiced. The EDR
archive CD-ROMS will be stored in a CD-ROM jukebox to allow unattended
- For each image:
- ) Decompress from JPEG to standard image format.
- ) Apply calibration to convert the image to physical
units. This calibration makes use of pre-launch and in-flight
calibration information to correct for variations in detector
response, interpolate bad pixels, and remove instrument noise. It
then applies knowledge of the exposure time, gain and offset state,
filter, and lighting geometry to produce a 32-bit image in physical
units, normalized to a reference lighting geometry.
- Register overlapping images to one another and to the highest
resolution USGS global image mosaic available as a reference. This
registration will be done automatically, to sub-pixel precision, using
a moving-window cross-correlation maximization algorithm. The
registration will occur in image coordinates rather than map-projected
coordinates to minimize the effects of errors in the SPICE
information; errors in the SPICE will result in mispositionings of
pixels on the planet's surface. Should the USGS mosaic prove to be of
insufficient resolution for HIRES mosaic registration, select portions
of the intermediate resolution LWIR data will be mosaiced and
registered to the USGS mosaic to then act as a base for the HIRES
- Compute the dynamic range of each mosaic and convert to 8-bit
form, recording the conversion factors to return to physical units.
Cosmetic adjustment of residual brightness level mismatches between
adjacent images will also be performed at this point, and recorded so
that conversion back to physical units will be unaffected.
- Map-project all mosaics. Experiments have shown that use of the
currently-available SPICE data results in image mismatches on the
order of 10 pixels. Refinement of the SPICE data will be done using
cross-correlation between overlap regions in adjacent images, between
a feature catalog of control points and the images, and between the
images and the base image to be produced by USGS.
Summary of Processing and CD-ROM Volumes to be Produced
Description Total Size CPU Usage (Months) Number of CD-ROMs
1/1024 HIRES polar mosaics 640 MB 0.4 1
HIRES strip mosaics 18 GB 12.5 28
Total ~19 GB 13 29