MER Analyst's Notebook Forum

Here are Jim Bell's responses:

SheeUn:

"The second question: how to calculate exposure time for RGB composite if they are all different for each filter?"

"In regards to the second one. Making a mosaic from 3 greyscale images doesn't work either. The panorama still needs a color balance/equalisation. I would like to know the method (and software, hope it isn't ISIS) by which imaging department of JPL creates pans from both MERs. In particular, how they deal with exposure differences between images featuring sky and ground, only ground, ground and parts of the rover."

Jim Bell:

The first thing we do is calibrate the images to radiance units (physical brightness units like Watts per meter squared per micron per steradian, etc.). Calibration is done by the Pancam team at Cornell University. Calibration "corrects" for exposure time differences between different images within the same sequence. Calibrated images are archived on the PDS as files with "RAD" or "IOF" in the file name ("RAD" stands for radiance; "IOF" stands for "I over F", where I is the radiance of the scene and pi*F is the radiance of sunlight incident on the scene (at the top of the martian atmosphere)--so the ratio I/F is approximately the reflectance of the scene). If you want more details, either I or the PDS folks here can send you copies of two papers published in 2003 and 2006 in the Journal of Geophysical Research that describe the cameras and calibration in LOTS more detail, and a third JGR paper from 2006 describing some of the challenges of generating color images from Pancam data.

We then mosaic the images at Cornell using some home-grown software that uses the altitude and azimuth information in each image label to place each part of the mosaic in the right position. The program uses a process similar to the auto-mosaiking software that is now common on commercial digital cameras or commercial programs like Photoshop. Our software is written in the "IDL" array processing language. It is not publicly available (we don't have the staff/time to support public software distributions), but if you are an IDL user I would be happy to discuss getting you a copy of the code to use informally...

Finally, sometimes I still have to manually "smooth" out brightness gradients/seams in the sky, using software like Photoshop, if a mosaic was taken over the course of many days or at many different times of day or dust conditions. This is purely an "artistic" step--a judgement call--used for generating the prettiest pictures, and designed to simulate what the sky might seem like if the images had all been taken at one time. If done, this is a tweak that is done at the end, to the JPEG or TIFF images--not something done to the actual raw or calibrated data. If you re-create mosaics using calibrated images released from the PDS, you might see some of the original seams in sky brightness that could not be corrected by calibration.

SheeUn:

"In press releases by MER science team images are always uniform in color and brightness. For example: Sol1162."

Jim Bell:

Yes, a good calibration of the individual images making up a mosaic will indeed result in a relatively seamless result!

SheeUn:

I also made an observation that in press-releases there are usually two renderings of the same image shot through the same filters: approximately true color and false color. But how can this be? False color is one that represents colors beyond visible spectrum. And there is, apparently, no such thing as "approximately" true color. There is just true color, i. e. RGB composite where red is L4, green - L5 and blue - L6; everything else - false color. Could you please comment on this as I cannot get feedback from MER team."

Jim Bell:

This is a big thorny topic, which I go into in some detail in the second 2006 JGR paper that I mention above (and the first URL below). The short answer is that "true color" has specific mathematical definitions in the colorimetry community/literature, and those turn out to be very specific (naturally) to color as viewed by humans on Earth. For example, "true color" renderings often make assumptions about the scene being illuminated by average sunlight at Earth and include a scattered light component from average Earth blue sky (and different components/assumptions for cloudy days, etc.). Obviously none of this applies at Mars, and so to be true to the mathematical definitions of the true color community, our attempts to modify that scheme for Mars are often referred to as "approximate true color". Maybe we should call it "Mars true color" or something else, who knows. The topic of "true color" on Mars is quite immature; unfortunately, none of us on the rover team have had time to develop it more fully.

False color is just that--a false rendering of the color of a scene, required when taking pictures through our filters that go farther into the UV or IR than human vision does. I am not sure what your concern is--think Andy Warhol and Campbell's Soup cans...

For more details on the particular issues that we deal with for "true" or false color renderings, here are some useful web sites:

http://marswatch.astro.cornell.edu/pancam_instrument/projects_1.html

http://www.cg.tuwien.ac.at/research/theses/matkovic/node9.html

http://www.badastronomy.com/bad/misc/hoagland/mars_colors.html

http://marsrovers.jpl.nasa.gov/spotlight/spirit/a12_20040128.html

I hope this information is helpful--sorry I don't have more time to get into a more detailed discussion on this. There are many online forums and resources, however--hopefully they will help too.

Have fun with the images!

-Jim Bell

Cornell U.