QUESTION:
What kind of compression software and algorithms did you use? What were your
constraining factors and what did you base your decision on?

ANSWER from Bridget Landry on July 15, 1997:
We have two main types of compression software, a lossy version (similar to
JPEG) and a lossless version (developed from the RICE algorithm).  With our
lossy compression we have many different options which include: Least Cosine
Transform, Arithmetic Encoding, Huffman tables, different quantization
tables, and pixel averaging, compression ratios and quality factors.  Some
of these options are used together for a single image.

Some of the constraining factors were:
1) Development time / delivery schedule.
2) Robustness to data loss.
3) Memory usage.
4) Image processing time.

Chris

ANSWER from Guy Beutelschies on July 25, 1997:
The pictures are sent back as pixel mapped binary data which is then Reed
Solomon encoded to increase the reliabilty of the link.  We also have the
option to compress the data using either JPEG or Rice compression
algorithms.

ANSWER from Chris Salvo on July 27, 1997:
There are a number of encoding, correction, and compression schemes at work
throughout the system.  A comprehensive answer would require a bit of
research on my part, as I am not completely familiar with the details of all
parts of the mission.
     
I will give you the highlights as I remember them off the top of my head,
and some hints about what I might be overlooking.
     
Highlights:
     
Deep Space Network / Lander Interface (Consult CCSDS - Consultative
Committee for Space Data Systems - for details):
     
- Uplink to the spacecraft uses 56 bit codeblocks with 8 bits (actually 7
plus a fill bit) of EDAC (error detection and correction).  The Hardware
Command Decoder on the spacecraft can correct single bit errors using these
bits.  There is also a checksum of some kind at the next level of
abstraction (what we call frames).
     
- Downlink from the spacecraft uses both Reed-Solomon encoding and a
convolutional code (Viterbi code, rate 1/6, constraint length 15).  The use
of these codes allows a lower signal to noise ratio for a given bit error
rate.
      
Lander internal communication:
      
- There is a 1553 bus onboard for command and data functions between the
Flight Computer and other pieces of hardware (the High Gain Antenna
Actuators, the Radar Altimeter, etc..  The protocol is standard - I believe
there is error detection and correction inherent, but I would have to check
on the details.
      
- There is also a VME backplane with its own inherent checking.
      
- The RAM storage is checksummed and has single bit correction and double
bit detection.  This is done in hardware not software.
      
Lander / Rover Comm:
      
- I am not that familiar with the details of this interface.  I do know that
it uses a handshaking mechanism and does detect errors.  I do not know if it
can correct some.
      
Imaging Data Compression:
      
- The imager on the lander (IMP - Imager for Mars Pathfinder) uses a wide
variety of compression algorithms.  It is capable of producing everything
from raw frames, to losslessly compressed frames, to lossy highly compressed
frames.  It can also subframe to crop the images as desired.  The
compression types include RICE, and LCT.  Pixel averaging (blocking) can
also be used alone or in combination with other compression.
      
Much of the imaging that was received on the ground in the first few days of
surface ops was compressed 12:1 or more.  The "gallery pan" which was
unveiled in D.C. last week was 6:1.  These are all lossy compression.
      
We are now working on the "super pan" which is losslessly compressed.
      
- The rover can also compress its images.  I am not familiar with the
algorithms involved.