DESCRIPTION |
Introduction to the Cassini Imaging Science Subsystem:
Narrow Angle Camera
Instrument Overview
===================
The Cassini ISS consists of two fixed focal length telescopes, a
narrow angle camera (NAC) and a wide angle camera (WAC). The NAC is
95 cm long and 40 cm x 33 cm wide, and has a focal length of 2002.70
+/- 0.07 mm in the clear filter. The two cameras together have a mass
of 57.83 kg, and sit on the Remote Sensing Palette (RSP), fixed to
the body of the Cassini Orbiter, between the Visual and Infrared
Mapping Spectrometer (VIMS) and the Composite Infrared Spectrometer
(CIRS), and above the Ultraviolet Imaging Spectrometer (UVIS). The
apertures and radiators of both telescopes are parallel to each other.
The NAC has its own set of optics, mechanical mountings, CCD, shutter,
filter wheel assembly, temperature sensors, heaters, and electronics,
the latter of which consists of two parts: the sensor head subassembly
and the main electronics subassembly. The Sensor Head electronics
supports the operation of the CCD detector and the preprocessing of
the pixel data. The Main Electronics provide the power and perform
all other ISS control functions, including generating and maintaining
internal timing which is synchronized to the Command Data System
(CDS) timing of 8 Hz, control of heaters, and the two hardware data
compressors. The Cassini Engineering Flight Computer (EFC) is a
radiation-hardened processor that controls the timing, internal
sequencing, mechanism control, engineering and status data
acquisition, and data packetization.
The NAC is an f/10.5 reflecting telescope with an image scale of ~6
microrad/pixel, a 0.35 deg x 0.35 deg field of view (FOV), and a
spectral range from 200 nm - 1100 nm. Its filter wheel subassembly
carries 24 spectral filters: 12 filters on each of two wheels. This
allows for in-line combinations of filters for greater flexibility.
Each wheel is designed to move independently, in either the forward
or reverse direction, at a rate of 3 positions per second. A homing
sensor on each wheel defines a home wheel position, and wheel
positioning can be commanded absolutely or relatively.
Unlike the WAC, the NAC is thermally isolated from the RSP in order to
minimize the effects of RSP thermal transients on the NAC image
quality.
The temperature of the CCD is controlled by a passive radiator,
directly connected to the focal plane, along with an active
'performance' heater on the CCD to adjust the temperature. The
temperature of the optical elements is controlled by active heaters
positioned along the optical path. These optical elements are kept to
within 1 degree Celsius to maintain camera focus without an active
focusing mechanism. Low expansion invar spacers are also used. The
radiator subassembly also includes two sets of spacecraft-controlled
decontamination heaters which are used to minimize deposition of
volatile contaminants on either the detector or radiator and to
minimize radiation damage to the CCD. All heaters are commandable (ON
or OFF) during flight.
Optics
------
The narrow angle camera optics were specially designed to improve on
the quality and resolution of images of the bodies in the Saturn
system returned by Voyager. It is based on a Ritchey-Chretien
reflector design. The focal plane field of view is limited by the
size of the CCD. The NAC point spread function (PSF) was designed to
be approximately the same physical size as a pixel in the near-IR.
The full width at half maximum (FWHM) of the PSFs of the NAC through
the clear filters is 1.3 pixels. The nominal pixel scale is 5.9907
microradians/pixel.
All the reflective optical elements within the NAC (the primary and
secondary mirrors) are manufactured of fused silica; all refractive
NAC elements (such as the field correctors and the window on the
sealed CCD package) are made of either fused silica or single-crystal
vacuum-UV-grade calcium fluoride. Antireflection coatings consisting
of single layer MgFl2 were deposited on the field correctors and CCD
window; a multi-layer MgFl2 coating was applied to the primary and
secondary aluminum-coated mirrors to enhance reflectivity. A fused
silica quartz plug is placed immediately in front of the CCD package
to protect the detector against radiation damage and to minimize
radiation- induced noise in the images.
Geometric fidelity in the NAC is very good: pre-flight analytical
calculations indicate distortions of less than a pixel at the corners
of the field of view, and subsequent observations of the Pleiades and
the open cluster M35 set the value to 0.45 pixels.
Filters
-------
The ISS filter assembly design -- consisting of two filter wheels and
a filter changing mechanism -- is inherited from the Hubble Space
Telescope WF/PC camera. Each wheel is designed to move independently,
in either the forward or reverse direction, at a rate of 3 positions
per second in the NAC. A homing sensor on each wheel defines a home
wheel position: wheel positioning can be commanded absolutely or
relatively.
The Cassini Imaging Science Team has deliberately duplicated 63% of
the filters in both the NAC and WAC. These include seven
medium/broadband filters from the blue to the near-IR for
spectrophotometry, 2 methane and 2 continuum band filters for
atmospheric vertical sounding, 2 clear filters, and a narrow band H
alpha filter for lightning observations.
The clear filter is in the 'home' slot of each filter wheel, since it
was deemed that sticking of a filter wheel, should it occur, was most
likely to occur in the home position. Typically a clear filter in one
wheel is combined with a color filter in the other wheel, though
two-filter combinations can also be used.
Because of its reflecting optics and its unique ability to see in the
UV, only the NAC carries filters for UV observations. The lumigen
coating provides a unique spectral capability, unavailable on either
the Voyager or Galileo imaging systems, which Cassini carries to the
outer solar system for the first time. It enables spectral response
down to 200 nm. To take advantage of this capability, we have spanned
the range from 230 nm to 390 nm with three UV filters: UV1, UV2, and
UV3.
The NAC filter wheel also contains narrow-band filters for atmospheric
studies. Methane absorption bands and continuum wavelengths are
available using the MT1/CB1, MT2/CB2 and MT3/CB3 filters. (CB1 is a
2-lobed continuum filter, with lobes on each side of the methane
absorption band.) A HAL filter is also included for observing H-alpha
emissions from lightning.
Finally, the NAC carries three polarization filters covering the
visible spectrum: P0, P60 and P120. As their names indicate, these
polarizers have principle transmission axes separated by 60 degrees,
in order to measure intensity and the degree and direction of linear
polarization regardless of camera orientation. The NAC also has a
single infrared polarizer, IRP0.
The polarizers are, of course, to be used in combination with other
spectral filters, so filter placement was important. In the NAC, the 3
visible polarizers and the one IR polarizer can all be used in
conjunction with a suite of spectral filters on the opposite wheel
covering the UV to the near-IR.
Table 1: ISS NAC Filter Characteristics
Filter Lambda_cen Lambda_eff Science Justification
----------------------------------------------------------------------
UV1 258W 264 aerosols
UV2 298W 306 aerosols, broadband color
UV3 338W 343 aerosols, broadband color,polarization
BL2 440M 441 medium-band color, polarization
BL1 451W 455 broadband color
GRN 568W 569 broadband color
MT1 619N 619 methane band, vertical sounding
CB1 619N 619 2-lobed continuum for MT1
CB1a 635 635
CB1b 603 603
RED 650W 649 broadband color
HAL 656N 656 H-alpha/lightning
MT2 727N 727 methane band, vertical sounding
CB2 750N 750 continuum for MT2
IR1 752W 750 broadband color
IR2 862W 861 broadband color; ring absorption band
MT3 889N 889 methane band, vertical sounding
CB3 938N 938 continuum for MT3, see thru Titan haze
IR3 930W 928 broadband color
IR4 1002LP 1001 broadband color
CL1 611 651 wide open, combine w/wheel 2 filters
CL2 611 651 wide open, combine w/wheel 1 filters
P0 617 633 visible polarization, 0 degrees
P60 617 633 visible polarization, 60 degrees
P120 617 633 visible polarization, 120 degrees
IRP0 746 738 IR polarization, see thru Titan haze
Table 2: NAC Two-Filter Bandpasses
Filters lambd_cen lambda_eff
--------------------------------
UV2-UV3 316 318
RED-GRN 601 601
RED-IR1 702 702
IR2-IR1 827 827
IR2-IR3 902 902
IR4-IR3 996 996
(All wavelengths in nm. Central wavelengths (lambda_cen) are computed
using the full system transmission function. Effective wavelengths
(lambda_eff) are computed using the full system transmission function
convolved with a solar spectrum. Bandpass types: SP = short wavelength
cutoff; W = wide; N = narrow; LP = long wavelength cutoff.)
With the exception of the clear filters and the polarizers, the
filters are all interference filters manufactured using an ion-aided
deposition (IAD) process which has the effect of making the filters
temperature and moisture tolerant, and resistant to delamination.
Conventional interference filters have passbands which shift with
temperature. The shift can be significant for narrowband filters
targeted to methane absorption bands or the H_alpha line. Temperature
shifts for IAD filters is typically an order of magnitude or more
smaller than for conventional filters and is insignificant over the
temperature range (room temperature to 0 degrees C) relevant to
calibration and operation of the Cassini cameras.
The NAC visible polarizers consist of a thin film (less than 1 microns
thick) of a polarizing polymer deposited between two fused silica
plates. The infrared polarizer has a 1 mm-thick layer of Polarcor
(trademark Corning) cemented between two slabs of BK7-G18 glass.
Polarcor is a borosilicate glass impregnated with fine metallic wires.
Ideal polarizers block only photons whose electric vector is
orthogonal to the principal axis of the polarizer. The visible
polarizers fall short of this ideal behavior in two ways. They
transmit too little of either polarization in the ultraviolet, and
too much of the light polarized orthogonal to the principal axis in
the near-infrared. Their performance is best between 450 nm and 650
nm where the principal axis transmission is between 0.45 and 0.65,
and the orthogonal transmission is less than 1%. The useable range of
the visible polarizers extends from the UV3 filter near 350 nm to the
CB2 filter at 750 nm. The infrared polarizer has much better
performance over its range (700 nm - 1100 nm) where the principal
transmission is greater than 0.9 and the orthogonal transmission is
0.001 or less.
Shutter
-------
Between the filter wheel assembly and the CCD detector is the shutter
assembly, a two blade, focal plane electromechanical system derived
from that used on Voyager, Galileo and WFPC. To reduce scattered
light, the shutter assembly was put in the optical train `backwards ,
with the unreflective side towards the focal plane. Each blade moves
independently, actuated by its own permanent magnet rotary solenoid,
in the sample direction: i.e., keeping the blade edge parallel to the
columns of the CCD. The shutter assembly is operated in 3-phases: open
(one blade sweeps across the CCD), close (the other blade sweeps
across the CCD to join the first), and reset (both blades
simultaneously sweep across the CCD in the reverse direction to the
start position).
There are 64 commandable exposure settings which can be updated during
flight if so desired. These correspond to 63 different exposure times,
ranging from 5 milliseconds to 20 minutes, and one `No Operation
setting. The shortest nonzero exposure is 5 msec. In the ISS flight
software, the time tag on the image is the time of the close of the
shutter. Because of mechanical imperfections in the shutter mechanism,
there is a difference between the commanded exposure time and the
actual exposure time, and a gradient in exposure time across the CCD
columns. At an operating temperature of 0 degrees C, the mean
differences in the NAC for commanded exposure times of 5, 25 and 100
ms were measured to be 0.98, 1.52 and 0.97 ms, respectively. In all
cases the actual exposure times are less than the commanded times.
There is also a small temperature dependence to these shutter offsets.
The 1024th column is illuminated first in both cameras. In the NAC,
this column is illuminated for ~ 0.3 msecs longer than the first
column. This value is independent of exposure time and reasonably
independent of temperature. The expected precision or repeatability
of an exposure (equal to the standard deviation of actual exposure
durations measured at any one location on the CCD in ground tests) is
= 0.03 msec for the NAC. Corrections for the mean and the
spatially-dependent shutter offsets are incorporated into the Cassini
ISS calibration software (CISSCAL). The shutters were tested for
light leak. None was detectable in the NAC at a fluence level of
12,000 times full well exposure on the closed shutter.
Detector
--------
The CCD detector used in the Cassini ISS was manufactured by Loral,
packaged by JPL, and employs three phase, front-side-illuminated
architecture. The imaging area -- the region on which light is focused
-- is a square array of 1024 x 1024 pixels, each 12 microns on a side.
The CCDs on both cameras were packaged, hermetically sealed and
fronted by a fused silica window.
The CCD's response to light is determined by the spectral dependence
of each pixel's quantum efficiency: i.e., the number of electrons
released in the silicon layer for each photon incident on it. In
front-side-illuminated CCDs (like that in the Cassini ISS), the
overlying polysilicon gate structures don t transmit UV light. To
achieve the required UV response, a UV-sensitive organic fluorescent
material called lumogen was vacuum-deposited onto the CCD at 80
degrees C after it was bonded. In this 0.6 micron layer, UV photons
are converted into visible photons in the 540 to 580 nm range that
readily penetrate the silicon below. Under vacuum conditions, the
lumogen layer would tend to evaporate when CCD temperatures reached
60 degrees C. For this reason, the CCD sealed packages were
back-filled with inert argon gas to a half atmosphere pressure. All
flight candidate CCDs were coated with lumogen before the two flight
CCDs were chosen and assigned to each camera.
The efficiency of a CCD in the near-IR depends on its thickness, or
more precisely on the thickness of the very thin, high purity silicon
layer which is epitaxially grown over a thicker (~500 micron)
substrate. It is the photons absorbed in the epitaxial layer that are
converted into the signal electrons that are subsequently collected
and sampled. Nearly all of the near-IR photons actually penetrate
beyond the epi layer and create charge in the substrate. However, the
purity contrast between the substrate and the epi layer prevents
substrate-generated charge from entering the epi layer and being
collected. Thus, the 1100 nm quantum efficiency is essentially the
fraction of incident flux which is absorbed in the thin layer of pure
silicon: the thicker the epi layer, the higher the infrared
sensitivity. However, the thicker this layer, the lower the spatial
resolution. A compromise was made in the manufacture of the CCD to
yield some response near 1100 nm while maintaining high spatial
resolution. The epi layer is 10 - 12 microns thick on Cassini and
results in a quantum efficiency (QE) of ~1% at 1000 nm.
A compromise involving the near-IR response was also made in choosing
the CCD operating temperature. At Saturn, this temperature is -90 +/-
0.2 degrees C and is a compromise between yielding an acceptably low
dark current (= 0.3 e-/sec/pixel) and maintaining a reasonable
near-IR response (which is diminished at low temperatures). CCD
thermal control is achieved by means of balance between passive
radiation to space, which alone would maintain the CCD below its
operating temperature at Saturn, and active heater control. The
radiator of each camera also supports a decontamination heater (35
watts in all) that can heat the CCD to +35 degrees C to reduce the
deposition of volatile contaminants on either the detector or the
radiator. (Because damage to the CCD due to cosmic rays can be
annealed at elevated temperatures, the CCD operating temperature
during cruise, when data were not being collected, was maintained at
0 degrees C to minimize such damage.)
The detector system includes an unilluminated region 8 samples wide -
the 'extended pixel' region - extending into the negative sample
direction in the serial register. These pixels get read out first.
Moreover, once an entire row of 1024 pixels is read up into the
serial register and out to the signal chain, the read-out continues
for 8 more clock cycles, or 'overclocked pixels,' to provide a
measure of the offset bias, the DN value that corresponds to zero
signal level. The extended pixel region and the overclocked pixels in
principle provide two independent measures of offset bias and a
sample of the horizontal banding pattern that may be used to remove
the pattern in images lacking dark sky. (A discussion of the
horizontal banding problem can be found in [PORCOETAL2004].)
In the NAC, the extended region of the readout register, and the
first 13 columns into the serial register are corrupted by a
grounding problem with the epoxy that bonds the pure silicon layer to
the substrate. This causes spurious swings in the voltage during the
initial 'clockings' of data out of the CCD into the signal chain.
Consequently, these columns of CCD data are unreliable, and the NAC's
extended pixel region cannot be used to monitor the camera's bias or
noise state.
Scientific Objectives
=====================
See [PORCOETAL2004] for an in-depth description of Cassini ISS science
objectives.
Camera Operation
================
Operational States
------------------
The ISS has three operational power states: On, Sleep and Off. In the
On state, the cameras are Active or Idle. In this state, both the
spacecraft replacement heaters and ISS decontamination heaters are
off. The camera software has active control over the performance
heaters to set appropriate operating temperatures for the optics and
CCD detector. The Active sub-state is entered to collect science data
as well as for calibration and maintenance activities. Command
execution in the active state includes science data readout, filter
wheel movement, shutter movement, activation of light flood and
calibration lamps, and other high power consuming activities. Idle is
a background state in which no commands are executing. When the
camera is in Idle, uploads can be processed, real-time and 'trigger'
commands can be accepted from the CDS, and macros can be stored. The
execution of any command sends the camera into the Active state. The
camera always returns to Idle state after completing a command
sequence. In the NAC, peak power consumption during active imaging is
26.2 watts.
The ISS Sleep state is a non-data taking low power state that is used
when no activity is planned for an extended period of time. During
this state, the sensor head and main ISS electronics are drawing
power, and the optics and CCD heaters are on to maintain operating
temperature limits. Spacecraft controlled replacement heaters are
off. The decontamination heaters may be used, if necessary. In Sleep,
the NAC consumes 22.3 watts.
In OFF, no power is drawn by the ISS. The spacecraft controlled
replacement heaters and ISS decontamination heaters may be turned on
when necessary. The replacement heaters keep the ISS within allowable
flight nonoperating temperature limits and the decontamination heaters
can be used to provide for CCD protection from the radiation
environment and from the condensation of volatiles. In this state,
the NAC consumes 8.4 watts.
Detector Modes
--------------
The CCD has the capability of being commanded to operate in full mode
(i.e., 1x1) or either 2x2 or 4x4 on-chip pixel summation modes. The
latter two modes are used for either enhancing signal-to-noise and/or
decreasing the data volume and/or read-out time at the expense of
spatial resolution. The full well of the CCD is roughly 120,000
e-/pixel. Four gain states are available: for imaging faint objects
(high gain, Gain 3) and bright objects (normal gain, Gain 2), and to
match the output of the 2x2 (Gain 1) and 4x4 (Gain 0) full wells. The
summation well can hold only 1.6 x 10^6 electrons; this corresponds to
full well with 4X4 summing. However, the relation between number of
electrons in the signal and the digital data numbers (DN) into which
the signal is encoded starts to become nonlinear above 10^6 electrons
because at this signal level, the on-chip amplifier becomes
non-linear. For this reason, in the lowest gain state (Gain 0), the
full scale signal is set to correspond to ~ 10^6 electrons at 4095 DN.
Table 3: NAC Gain States
Gain State e-/DN Notes
----------------------------------------------------------------
0 233 +/- 29 Designed for 4x4 summation mode
1 99 +/- 13 Designed for 2x2 summation mode
2 30 +/- 3 Normal gain; used in 1x1 summation mode
3 13 +/- 2 Used in 1x1; chosen to match read noise
The capability also exists within the ISS to reduce the effect of
blooming, the phenomenon whereby a highly overexposed pixel can spill
electrons along an entire column of pixels, and sometimes along a row,
when the full well of the CCD is exceeded. The default camera setup
has anti-blooming on, with the option to turn it off. Anti-blooming
mode is achieved by applying an AC voltage to the chip, forcing
excess electrons into the silicon substrate. An undesirable side
effect of this action is to pump electrons into traps in the silicon
at the expense of electrons in adjacent pixels. For long exposures
this produces bright/dark pixel pairs. These were initially present
in nearly all the NAC flat field files obtained during calibration in
the thermal vacuum chamber. Corrected flat field files with these
pixel pairs removed have since been created.
Camera Commanding
-----------------
The acquisition of images can be accomplished in several ways.
Individual NAC or frames may be acquired, or the NAC and WAC can be
used in simultaneous mode, called BOTSIM (for 'both simultaneous').
The entire event, which is called a framing event and requires a
total duration called a 'framing time', is broken down into two
steps: the prepare cycle and the readout cycle.
The prepare cycle is used to alter the state of the ISS, step the
filter wheels, perform heater operations, light flooding, and other
functions required to prepare for an exposure. It also includes the
exposure time. The prepare cycle is constructed from a series of
quantized windows of time in which specific functions are assigned to
occur.
During the prepare cycle, the shutter blades are reset from the
previous exposure and the filter wheels are moved into position.
Because simultaneous motion of each filter wheel requires more power
than the ISS was allocated for peak operation, all filter wheels NAC
and WAC -- are moved separately. Windows of quantized duration are
set aside for the motion of each filter wheel. Next, the CCD is
prepared for exposure to light. This preparation begins with a wait;
the duration of the wait is chosen to ensure that the shutter will
close exactly at the end of the prepare cycle. After the wait, a
light-flood fills the wells of the CCD to many (~ 50) times
saturation, followed immediate by a read out. The entire
light-flood/erase event takes 950 msec and has the effect of erasing
any residual image of previous exposures from the CCD. Within 5 msec
of the end of the light-flood/erase event, the shutter is opened for
the commanded duration. (For dark frames, this duration is set to
zero.) The image is tagged with the time of shutter close.
During a BOTSIM, the prepare cycle is lengthened to include time to
prepare both NAC and WAC. The NAC is prepared first; then the WAC is
prepared so as to avoid simultaneous movement of any of the 4 filter
wheels. If the NAC and WAC exposure times are different, the exposures
begin in a staggered fashion so that the NAC and WAC shutters are
closed simultaneously. There are 63 discrete commandable exposure
times which are accommodated within 13 discrete prepare cycle windows.
During the following readout window, the CCD is read out, the data are
encoded and/or compressed, and the results are packetized. For any of
the 6 individual CDS pickup rates, there are 4 discrete readout
windows for each camera. The readout window is scaled by the CDS
pickup rate giving 24 actual readout windows per camera and 96 actual
BOTSIM readout windows.
Prepare times and readout times are chosen before uplink. The prepare
cycle is completely determinate; the readout time required to fully
read out an image is not. The required readout time during the image
event will depend on the amount of data being read out of the CCD,
and the CDS pickup rate or on the line readout rate from the CCD,
whichever is slowest. If the data volume in the image was
underestimated and the required readout time exceeds the commanded
readout time, the camera will cease reading out part way through an
image and lines will be lost. For this reason, a great deal of effort
has gone into the amount of data returned for different scenes and
choices of compression parameters.
The ISS can collect pixel (image) data, engineering data and status
data, and packetize them with appropriate header information as either
science telemetry packets (which include all types of data) or
housekeeping packets (which only include engineering and status data).
The latter are returned alone when ISS is in an ON power state but not
actively taking images. The frequency with which housekeeping packets
are collected is 1 packet/sec and is programmable in flight. The
amount of housekeeping data that gets sent to the ground is
determined by the rate at which CDS picks up such packets and is
currently 1 housekeeping packet every 64 seconds.
Data paths
----------
The analog to digital (A/D) conversion happens right as the analog
signal is read out from the chip, after it has passed through the
on-chip amplifier. Data from the ADC are encoded to 12-bit data
numbers (DN), giving a dynamic range of 4096. However, they are
stored as 16-bits: the upper 4-bits are all 1 s. The ISS flight
software masks the upper 4 bits when doing calculations. Compression
and conversion functions are performed after the electrons are
converted to DN. The next juncture is a choice of data conversion
(from 12 to 8 bits) or no data conversion. Unconverted data can then
proceed to a lossless compressor or undergo no compression at all.
Converted data can undergo no compression or lossless or lossy
compression. From there, the data are placed on the Bus Interface
Unit (BIU), where they are ultimately picked up by the Command Data
System (CDS) and sent to the Solid State Recorder (SSR) where they
are stored as 16-bit data.
Data Compression
----------------
Serious constraints are imposed on imaging of the Saturn system by the
limited storage volume on the spacecraft's SSR, and by the limited
communication bandwidth back to Earth. In order to make the most
effective use of these resources, the Cassini imaging system includes
the capability to convert the data from 12 bits to 8 bits (called data
conversion), and also to perform either 'lossless' or 'lossy' image
compression. Data conversion, and both lossless and lossy compression,
are implemented in hardware.
Conversion to 8 bits
--------------------
Two sub-options are available for 8-bit conversion. One is a variant
on conventional 'square root' encoding. In such encoding, a look-up
table (LUT) is used to convert the original data values to 8-bit
values. The output 8-bit values are related to the input values in a
non-linear fashion, typically scaling with the square root of the
12-bit value. This non-linear scaling more closely matches the
quantization level to the photon shot noise so that the information
content is spread more evenly among the 256 levels. (The Cassini
12-to-8 bit conversion table is provided with the calibration data
volume.) It differs somewhat from pure square-root encoding, having
been designed for the known noise properties of the Cassini cameras
to distribute quantization-induced errors uniformly across the
dynamic range of the system. The look-up table is stored in ROM
within the cameras' memory and cannot be altered in flight; choice of
ON or OFF is commandable in flight.
The other sub-option is conversion from 12 bits to the
least-significant 8 bits LS8B). This type of conversion is useful for
reducing the data volume of images taken of very faint targets, such
as diffuse rings or the dark side of Iapetus, which generally do not
yield large signal levels and can be encoded to the lowest 8 bits.
Lossless Compression
--------------------
Both converted (8-bit) and unconverted (12-bit) data can be lossless
compressed. The ISS lossless hardware compressor is based on Huffman
encoding, a high efficiency, numerical encoding scheme in which the
length of the bit sequence used to encode a given number is chosen
based on the frequency of occurrence of that number. In ISS lossless
compression, each compressed image can be reconstructed on the ground
with no loss to the information content of the image, provided the
image entropy does not exceed the threshold where 2:1 compression is
achieved. Scenes with low entropy will have compression ratios higher
than 2:1; scenes with high entropy will never compress greater than
2:1, but the ends of lines will be truncated so that the total amount
of data returned in a pair of lines of the image never exceeds the
total number of bits for a single uncompressed line. The truncation
scheme has been designed so that the truncation alternates -- i.e.,
every other line -- from one line to the next, on the right (large
sample number) side of the image. If the data loss is great, it can
in principle result in the complete loss of every other line. In
either case, with this scheme, information (though reduced in spatial
resolution) can be retained across the image.
Lossy Compression
-----------------
Imaging sequences requiring larger compression ratios than can be
achieved with the lossless compressor may instead be more strongly
compressed using the camera's lossy compression circuitry. This
capability requires that the data have been converted to the 8-bit
form. Consequently, data conversion must be employed first before the
data are sent to the lossy compressor. Compression is implemented by
a pair of specialized signal processing chips which perform a
variation on the familiar JPEG (Joint Photographic Experts Group)
compression algorithm used in many image transfer and storage
applications. The JPEG algorithm operates by selectively removing
information from an image, particularly at high spatial frequencies.
Lossy-compressed images thus tend to have reduced detail on fine
scales.
For More Information
====================
More information regarding the camera design, operation, imaging and
compression modes, and image calibration can be found in
[PORCOETAL2004]. Additional discussion of calibration can also be
found in the documentation for the calibration volume of this data
set.
|