Home > Archive > Compression > February 2005 > Selecting Wavelets for Compression : suitable for DSN images, 12 bits of depth, size
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Selecting Wavelets for Compression : suitable for DSN images, 12 bits of depth, size
|
|
| Max Power 2005-02-04, 3:55 am |
| The Mars Exploration Rovers "Spirit" (MER-A) and "Opportunity" (MER-B)
arrived safely on Mars on January 4, 2004, and January 25, 2004 (Universal
Time), respectively. Onboard image compression is used extensively to make
best use of the downlink resources.
Most of the images were compressed with the ICER image compression software.
The remaining images that were compressed made use of modified Low
Complexity Lossless Compression (LOCO) software.
ICER is a wavelet-based image compressor that allows for a graceful
trade-off between the amount of compression (expressed in terms of
compressed data volume in bits/pixel) and the resulting degradation in image
quality (distortion). When the compressed data volume is allowed to be large
enough, ICER will produce lossless compression.
The development of ICER was driven by the desire to achieve stateof-the-art
compression performance while meeting the specialized needs of deep-space
applications. In particular, ICER incorporates a sophisticated
error-containment scheme to limit the effects of data losses seen on the
deep-space channel.
ICER also features progressive compression: compressed information is
organized so that as more of the compressed data stream is received,
reconstructed images of successively higher overall quality can be
reproduced.
When lossless compression is desired, the MER mission generally uses a
software implementation of a modified version of the LOCO image compressor.
Although ICER can also perform lossless compression, the simple predictive
approach used by LOCO is several times faster, with similar compression
effectiveness.
In addition to conventional image compression, MER is using a handful of
other techniques to reduce image data volume.
These include sending only subframes of interest of certain types of images
(while possibly sending the whole image at much lower quality); performing
pixel averaging to reduce the size of images; companding images from 12
bits/pixel to 8 bits/pixel before compression; and sending only row sum
data, column sum data, or histogram data.
Also, for each image acquired, the rovers produce a small 64 × 64 pixel
"thumbnail" image by pixel averaging; these thumbnail images are compressed
by ICER prior to transmission. The thumbnails serve as previews of the
full-sized images, which may not be transmitted as soon due to the larger
data volume they occupy.
The statistics presented in this article encompass all images downlinked to
Earth as of about February 7, 2004. We do not include data produced by any
instruments other than the cameras, and we have omitted various small
one-dimensional image data sets for which ICER and LOCO are not applicable.
The rates and compressed data volumes presented here do not include
packetization overhead.
The images acquired by the rovers are of a variety of targets and have a
variety of purposes. Special types of images include images of a calibration
target, Sun-finding images, and the descent images. Naturally, a large
portion of the images have been of the Martian landscape, for various
navigation and science purposes.
=========================
Preliminary Image Compression Results from the Mars Exploration Rovers
A. Kiely and M. Klimesh
=========================
> Compression is achieved in many methods such as predictive coding and
> transformation. On what basis Wavelet transformation based compression
> is choosen. What characteristics is seen on the image for selection of
> wavelet transformation based compression.
|
|
|
|
|