IPB

Welcome Guest ( Log In | Register )

40 Pages V  « < 38 39 40  
Reply to this topicStart new topic
Juno development, launch, and cruise, Including Earth flyby imaging Oct 9 2013
mcaplinger
post Jun 21 2016, 08:08 PM
Post #586


Senior Member
****

Group: Members
Posts: 1560
Joined: 13-September 05
Member No.: 497



QUOTE (scalbers @ Jun 21 2016, 10:36 AM) *
I wonder if there is a reason for the greenish cast with these images...

If these are from the EDRs, then it's because there is no color correction in the camera and it just worked out this way between the filter bandpasses and the sqroot encoding. If they were from the RDRs with the band scaling applied, it's because the scaling is slightly off, which would not surprise me.

For the EFB images, since the Earth's clouds are basically white, an auto white balance works fine, though a color purist might object.


--------------------
Disclaimer: This post is based on public information only. Any opinions are my own.
Go to the top of the page
 
+Quote Post
Gerald
post Jun 21 2016, 10:42 PM
Post #587


Senior Member
****

Group: Members
Posts: 1765
Joined: 7-December 12
Member No.: 6780



The images of the above synopsis are rendered from the EDRs, and without additional gamma correction, so gamma = 2.0. I've derived a color scaling from the EFB01 Moon image. But I've done it more or less manually. So it might not yet be perfect. I might refine the weights later, after measuring Moon's color accurately. The images are also not yet flat-fielded; I've no explicite flat-field, and didn't yet try to derive a flat field; this might shift colors slightly, too. And there might be some smear and stray light contributing to a color cast; this latter effect will be much less at Jupiter. Then, small shifts in perspective seem to be sufficient to change the apparent brightess of the target and hence of the color bands, resulting in color casts. For colors which are dark in at least one band might be sensitive to small inaccuracies of dark current subtraction and other image noise.
Next, I didn't use a full 3x3 matrix linear color correction, but only use the main diagonal, i.e. linear factors for each band separately. Then, the color band spectra are different from human color receptor sensitivity spectra. The color filter characteristics might change with light incidence angle; this effect seems to occur at least for the CH4 band, if I interprete the respective section in the JunoCam paper correctly.
Another effect might be the slightly greenish cast of sunlight outside Earth's atmosphere, but I'd think, that this effect should cancel out with Moon's color.
So, using Earth as an inflight color-calibration target, instead of Moon is principally an option, with the above limitations regarding various sources of color casts. Necessary would be a color-correct Earth image of exactly the same time and perspective to infer the best color correction matrix.
The easier way is deriving the weights by comparing EDRs with linearized RDRs, to get consistent with the MSSS calibration, or using RDRs directly, with the constraints, Mike mentioned.

I had also a short discussion with astronomers about using telescopic Jupiter images as a color reference. But it seemed, that there doesn't exist a consensus how to color-calibrate telescopic Jupiter images.
My focus thus far has been - and will stay for some more time - improving geometric calbration.
So, I'm open to suggestions, how to improve color calibration - or about which image data are required by others to improve the images I'll hopefully be able to provide.
Go to the top of the page
 
+Quote Post
scalbers
post Jun 21 2016, 11:58 PM
Post #588


Senior Member
****

Group: Members
Posts: 1245
Joined: 5-March 05
From: Boulder, CO
Member No.: 184



Interesting info. My TPS blog post (including the comments) linked 3 posts back mentions a number of things about handling the Earth's color. In brief, I'd suggest the brightest clouds should be pretty close to white. The clear oceanic areas should approximate reasonably the blue sky as seen from the ground (at least at a low phase angle with clean air). More on the latter is here in terms of CIE color matching functions and the 3x3 matrix: http://markkness.net/colorpy/ColorPy.html

Maybe a Jupiter image (or photometry) from a telescope can be used as a second check - if the color shift due to atmospheric extinction is corrected for.


--------------------
Go to the top of the page
 
+Quote Post
Gerald
post Jun 22 2016, 10:53 AM
Post #589


Senior Member
****

Group: Members
Posts: 1765
Joined: 7-December 12
Member No.: 6780



Thanks, those are great treatments of the subject. I've read your TPS article shortly after it has been published. As soon as I'll be on a satisfying level with the geometric calibration, I'll go deeper into the color calibration detail, since I think, that proper understanding of the colors is crucial for understanding Jupiter's aerosols and atmosphere dynamics.
Accurate geometry is necessary to obtain - among other things like cloud topography and wind velocity - useful color data on the pixel scale.
Go to the top of the page
 
+Quote Post
Bjorn Jonsson
post Jun 22 2016, 11:53 PM
Post #590


IMG to PNG GOD
****

Group: Moderator
Posts: 1857
Joined: 19-February 04
From: Near fire and ice
Member No.: 38



When making color composites of Jupiter I usually adjust the color balance to make the biggest bright zones (e.g. the North Tropical Zone) roughly white near the center of the disc. Or if I'm processing a hi-res image (e.g. a Voyager closeup) where no such zone is visible I either use the same processing parameters I used in a more distant image where a zone was visible or I make one of the bright ["white"] ovals roughly white. I sometimes also use Jupiter's global spectrum at visible wavelengths as a guide. However, true planetary colors are highly subjective.
Go to the top of the page
 
+Quote Post
Gerald
post Jul 2 2016, 11:48 AM
Post #591


Senior Member
****

Group: Members
Posts: 1765
Joined: 7-December 12
Member No.: 6780



QUOTE (scalbers @ Jun 22 2016, 01:58 AM) *
... In brief, I'd suggest the brightest clouds should be pretty close to white..

Since the 2-fold supersampled JunoCam images can be rather large, I've registered a domain, together with some webspace.
Here, on this still somewhat rudimentary website, some of the (large!) supersampled reprojections of EFB05 to EFB13. I tried to adjust the color band weights a bit to get the clouds closer to white.
I'll hopefully be able to improve RGB band registering further, before the first hires Jupiter images will become available after Perijove 1.
Go to the top of the page
 
+Quote Post
Gerald
post Jul 3 2016, 11:10 AM
Post #592


Senior Member
****

Group: Members
Posts: 1765
Joined: 7-December 12
Member No.: 6780



QUOTE (mcaplinger @ Jun 20 2016, 09:54 PM) *
I have updated six of the EFB images at http://www.msss.com/junocam_efb/pds/ that were decompressed incorrectly. These are 00C096, 00C098, 00C100, 00M095, 00M099, and 00M103. The rest of the EFB dataset should have been correct...

While checking my reprojections for consistence by combining RGB-CH4-pairs to false color images, I've noticed, that 00M097 seems to show some decompression artifacts, too.
Go to the top of the page
 
+Quote Post
Gerald
post Jul 9 2016, 04:12 PM
Post #593


Senior Member
****

Group: Members
Posts: 1765
Joined: 7-December 12
Member No.: 6780



Emily has been so kind to host some of my processed JunoCam data on the TPS server, and to upload an animation on youtube.

In the meanwhile, among other things, I've fixed a constraint of my reprojection software, which added artifacts when reprojecting into the past.
Now I can reproject all dayside efbs to a common instant:
Attached Image
Attached Image

Attached Image
Attached Image


Attached Image
Attached Image

Attached Image
Attached Image


Attached Image


This is for several purposes, like merging of several swathes, feature tracking, and geometric calibration.
Go to the top of the page
 
+Quote Post
Gerald
post Jul 19 2016, 04:35 PM
Post #594


Senior Member
****

Group: Members
Posts: 1765
Joined: 7-December 12
Member No.: 6780



A pinhole-reprojected EFB12, 8000x8000 pixels (about 31 MB png).
Thumbnail:
Attached Image

I've introduced a small Brownian K2 to improve the registering in the overlap region a bit. But for the x/z scale I've an unresolved inconsistency of about 0.5% to obtain a good result for the overlap region vs. good result for the other parts of the image. So this version is still a compromise.
I've tested for an aberration due to a possible minor rotation of the camera around the z-axis with respect to Juno's rotation axis, but it's probably less than 0.1 degrees; I failed to resolve the small inconsistency that way. Moving the optical axis along the y-axis with respect to the nominal value didn't result in significant improvements either. Moving the optical axis a few pixels along x can help to improve registering the area near Earth's "left" limb.
...But now I'm curious about the Jupiter approach images, and hence uploaded the efb12 version as it is; I'd think, that another improvement of accuracy by a factor of about 5 to 10 should be possible. I may continue the EFB work later, which is mainly for calibration and test purposes, to prepare for PJ (perijove) image processing. Earth is better-suited for calibration purposes than Jupiter due to the much lower wind velocities here on Earth. On Jupiter, it is one objective to determine wind velocities, based on a good reference calibration.
Go to the top of the page
 
+Quote Post

40 Pages V  « < 38 39 40
Reply to this topicStart new topic

 



RSS Lo-Fi Version Time is now: 25th April 2017 - 08:31 PM
RULES AND GUIDELINES
Please read the Forum Rules and Guidelines before posting.

IMAGE COPYRIGHT
Images posted on UnmannedSpaceflight.com may be copyrighted. Do not reproduce without permission. Read here for further information on space images and copyright.

OPINIONS AND MODERATION
Opinions expressed on UnmannedSpaceflight.com are those of the individual posters and do not necessarily reflect the opinions of UnmannedSpaceflight.com or The Planetary Society. The all-volunteer UnmannedSpaceflight.com moderation team is wholly independent of The Planetary Society. The Planetary Society has no influence over decisions made by the UnmannedSpaceflight.com moderators.
SUPPORT THE FORUM
Unmannedspaceflight.com is a project of the Planetary Society and is funded by donations from visitors and members. Help keep this forum up and running by contributing here.