Process JunoCam Raw with PI?

[f11]

In a day or so NASA will release the raw image data from the Juno probe of Jupiter's great red spot, taken July 10th on its recent flyby - it was REALLY close, only some 5800 miles above it.  I was wondering if anyone here has developed a PI script to process the raw image - a 1648 pixel wide by 384xn pixel high "strip" where each of the n 384-pixel high frames contains 3 x 128 pixel high blue, green and red filtered mono images, or "framelets" - into a single colour-combined image.

The only unusual part for PI is separating the n x 384 pixel high frames from the raw image "strip" into separate image files, then sub-dividing those frames into three 1648 pixel wide by 128 pixel high "framelet" images to separate the colours.  After that, you have to stitch together the same-colour framelets to create a set of three full sized filtered images.  Then, standard PI colour combination processing can be done.

There are some purpose-built programs to process these images (and do a lot more as well) like SPICE and ISIS3, but after looking at them, they're more than I want to learn for now.  I can do all the steps manually using various tools, but creating finished 3-colour images of NASA-supplied raw image data from space-faring probes seemed like it might fit into the fringe of PI's baliwick.

I was just starting to fiddle with PixelMath to see if the necessary functions exist to automate separation of the framelets and stitching them together when it occurred to me somebody here may have already done it.  Anyone?  Or is my application just a little too far outside the design parameters of PI?

Failing that, has anyone here developed a python or perl script to do bulk frame/framelet separation and stitching tasks?  Any suggestions welcome.

[drmikevt]

I'm not quite sure what I'm looking at yet, but here is the link to the raw data:  https://www.missionjuno.swri.edu/junocam/processing/

[drmikevt]

I looks like it is already separated into R,G,B images.  Combing them with Channel Combine seems to work well, and then applying curves gets you some nice color.  Cool!

[georg.viehoever]

The data you can get on https://www.missionjuno.swri.edu/junocam/processing/ is already preprocessed. To quote the website:

"Junocam pixels are 12 bits deep from the camera but are converted to 8 bits inside the instrument using a lossless "companding" table, a process similar to gamma correction, to reduce their size.  All Junocam products on the missionjuno website are in this 8-bit form as received on Earth.  Scientific users interested in radiometric analysis should use the "RDR" data products archived with the Planetary Data System, which have been converted back to a linear 12-bit scale."

In addition to that, NASA apparently dropped the "methane" channel, and they aligned the stripes to each other (probably knowing the exact movements of the Juno and Jupiter relative to each other - something we dont know). The "RAW" data provided omits channel extraction and alignment, so you can see the stripes as received on earth. If you wan to experiment with the data, I would recommend to start with the provided RGB version.

If you still would like to work with this "RAW" data: PixelMath can do it, except for the stripe alignment. The necessary expressions and symbols can be seen in the attached screenshots. In Image30 you can clearly see that there are still offsets between the stripes, and I found no way to fix these by means of some simple expression. Maybe the PixelMath Gurus in this forum have some brilliant idea.

Enjoy!
Georg

[georg.viehoever]

Still, a lot can be done with the data that is available. A quick try with the image that I downloaded (left) shows that you can nicely enhance color and details (right). Unfortunately, this also enhances the flaws in the NASA data, such as dead pixels, noise, alignment imperfections, seams between stripes. See screeshot.

Just give the NASA data a try. It is fun to play with it!

Georg

[drmikevt]

Yes - this processing took about 2 minutes.  I'm sure others could do a better job - I have no experience at all with planetary data.  How do you color calibrate, anyway?  I've just used curves and a few contrast enhancement processes.

[f11]

Thanks all for your input.  I was traveling most of today so didn't see all these responses until an hour ago.

Yup, I know NASA provides the RGB assembled frames, but I thought I'd try to work with the raw data at least once before taking the easier RGB route