Space Topics: Space Imaging
Image Formats and Metadata
[EDITORIAL NOTE: A discussion of FITS, JPEG, GIF, PNG, and JPEG2000 formats
needs to be added to this page.]
The images sent back from space missions are data -- the brightness or darkness
of each pixel in an image represents a precise measurement of the intensity
of light that is reflected or emitted by some object of interest to planetary
scientists. Therefore, it is critically important to researchers that
the value of every pixel in a planetary image be unaffected by the way it is
stored. It is also important that every image be associated with metadata
-- literally, "data about data." Image data from planetary
missions would be nearly useless unless they were accompanied by such information
as when it was taken, by what spacecraft, with what instrument, pointed at
what target, and so forth.
VICAR Format
Image data from the earliest planetary missions that returned imaging data
(e.g., Mariner 9 to Mars, Mariner 10 to Venus and Mercury, Viking Orbiter
and Lander missions to Mars, and the Voyager mission to the outer planets)
was processed at the Jet Propulsion Laboratory (JPL) in the 1970s and 1980s. At
that time, almost all of the data formats now in routine use did not exist. All of the images from these early missions were
processed using a suite of image processing programs known as VICAR (for “Video
image access and retrieval”). The final image data products were
sent to the data archives in VICAR format. VICAR format includes a
file header (usually ASCII text format) containing the metadata and image
data stored line by line in individual records as binary data.
In the 1970s and early 1980s, the primary storage medium used to archive planetary
image data was magnetic tape. After several years, physical deterioration
of the tapes was observed. In addition, the computers and operating systems
originally used to store the data onto the tapes were becoming obsolete. In
the mid-1980s, NASA funded JPL to recover as much as possible of the image
data from the original tapes and archive the data onto contemporary media. The
data recovery task produced CDs containing as much data as could be recovered
at that time. The data format was not changed during the preservation
project, which means that data from missions archived during the data preservation
task are still in VICAR format. Internet access to the CD-based data
sets is provided by the PDS for those missions. Public domain software
exists to read the data and convert it into data formats more commonly in use
by today’s commercially available image processing software (click here
for a list of software useful for opening vicar formatted data).
PDS Format
During the 1980s, the Planetary Data System established standards governing
the format of image data delivered to the PDS. Later missions conformed
to that format standard, and the PDS provides online software enabling users
to download data from the PDS archives and convert it into other useful data
formats. PDS formatted images usually have a .IMG file extension and have detached
ASCII-formatted text labels containing the image's metadata.
"Raw" Formats
The word "raw" is currently used in two contexts to describe spacecraft
images. Recently, as the Internet has become widely accessible to the
general public and standard image formats have evolved, several NASA planetary
missions, starting with Mars Pathfinder, have made mission data available via
the Internet within days or even hours of the receipt of data from the spacecraft. These
early versions of the images are often called "raw" images. They
are stored in lossy formats (such as JPEG), have typically not been calibrated
by their instrument teams, and are usually not accompanied by much metadata,
so they are not useable for scientific research. They are also usually "stretched" to
enhance their contrast, which makes it impossible to generate "true color"
images from them. They
are released on the Web only to permit the public to watch the current status
of a mission through a spacecraft's eyes.
"Raw" is also used to describe image data that is represented as
it was returned from the spacecraft, before any calibrating or processing has
been performed to correct for idiosyncrasies of the instrument, geometry, lighting
conditions, etc. Vidicon cameras especially produced images with noticeable
geometric distortions and brightness variations, but even images taken with
modern CCD cameras must be calibrated before they are scientifically useful. Each
instrument's science team processes, corrects, and verifies its own image products. The
PDS receives both the initial releases of raw image products from current missions
and the final scientifically corrected versions. ESA's Planetary Science
Archive operates similarly.
Archival data records of planetary image data are available at a variety of "levels." There
are standards that define the various levels of data processing that can be
performed on planetary image data. The standards were developed by a
National Research Council Committee on Data Management and Computation (CODMAC)
in the 1970s. The standards are shown in the following table.
CODMAC
Data Product
Level |
CODMAC Definition |
0 |
Raw Data. Data set corrected for telemetry errors
and decommutated. Data are tagged with time and location of acquisition. |
1A |
Edited data. Unresampled data that are still
in units produced by the instrument, but have been corrected so that
values are expressed in, or are proportional to, some physical unit. (For
example, images that have been flat-fielded to remove blemishes.) |
1B |
Resampled data. Data that have been resampled
or reprocessed in such a way that the original edited data cannot be
reconstructed. (For example, images that have had geometric distortion
removed.) |
2 |
Derived data products containing geophysical variables
at the same resolution and location as the Level 1 source data. (For
example, color images or anaglyphs produced by overlaying images without
any geometric projection.) |
3 |
Variables mapped on uniform space-time grid scales, usually
with some completeness and consistency. (For example, images that have
been map-projected and/or mosaicked.) |
4 and above |
Model output or results from analyses of lower level
data (For example, maps of mineral abundance derived from spectral data.) |
Many thanks to Bill Green for his help in developing this
page.
|