search

SETI / rms-opus

PDS OPUS - Outer Planets Data Search Tool
Apache License 2.0
10 stars 7 forks source link

Cassini ISS previews are too dark #254

Closed rfrenchseti closed 5 years ago

rfrenchseti commented 6 years ago

https://tools.pds-rings.seti.org/opus#/instrumentid=Cassini+ISS&timesec1=2017-001T00:00&camera=Narrow+Angle&FILTER=CLEAR&target=S+RINGS&projectedradialresolution2=1&view=browse&browse=gallery&colls_browse=gallery&page=1&gallery_data_viewer=true&limit=100&order=time1&cols=ringobsid,planet,target,phase1,phase2,time1,time2&widgets=projectedradialresolution1,FILTER,camera,timesec1,instrumentid,planet,target&widgets2=&detail=

From Michael Aye Michael.Aye@lasp.colorado.edu Thanks, that’s all what I need for now! Seems to work for me.

The only thing I’m wondering is, why many of the ISS preview’s are so dark. See: https://tools.pds-rings.seti.org/opus#/instrumentid=Cassini+ISS&timesec1=2017-001T00:00:00&timesec2=&camera=Narrow+Angle&FILTER=CLEAR&target=S+RINGS&projectedradialresolution1=&projectedradialresolution2=1&view=browse&browse=gallery&colls_browse=gallery&page=1&gallery_data_viewer=true&limit=100&order=time1&cols=opusid,planet,target,phase1,phase2,time1,time2&widgets=projectedradialresolution1,FILTER,camera,timesec1,instrumentid,planet,target&widgets2=&detail=

But I think that’s out of your control, as those are produced by the ISS team, IIUC. Anyhow, my tools are downloading and recalibrating those anyway, so I am fine, but others might find it odd, too.

Best, Michael

rfrenchseti commented 5 years ago

[From Mark]

Colleagues,

I checked and, to my surprise, the COISS preview generator does not use any percentiles when determining the stretch. That explains the prevalence of dark images--the brightest image in the image, even if a cosmic ray hit, defines the stretch.

The problem is more serious than that, however. I noticed that we have a lot of images like this:

n1488897368_1

If I choose any percentile at all, then Mimas gets washed out. So percentiles don't work.

I came up with a new option in picmaker (not yet checked in) where you can apply a 2-D median pre-filter across the image before calculating the limits. Using a 5x5 box, I get good results for Mimas above and also for the image Rob pointed out:

n1868838571_1

HOWEVER, as Rob also pointed out, we will have a problem with the median filter approach if applied to star fields, because then the median filter will wipe out all the stars and over-stretch the image to the extreme.

So this is an open research problem. At some point we will need to regenerate all the previews (not just COISS) using a better approach. However, finding a method that works well on small moons, large rings, and star fields is going to be challenging. Suggestions welcome!

--Mark

rfrenchseti commented 5 years ago

Kamino closed and cloned this issue to SETI/pds-webserver