Cory Hughes
New Member
Posts:6  
| 02 Dec 2019 06:36 PM |
|
|
Hello,
I am working with Mars Reconnaissance Orbiter - Compact Reconnaissance Imaging Spectrometer for Mars data. It is a spectrometer with ~438 band passes that image a relatively small footprint of the martian surface at ~18 meters/pixel.
I am trying to extract mean spectra from regions of interest (ROIs) for analysis of the minerals on the surface of the Red Planet to help inform us about when Mars may have been most hydrologically active.
The issue that I am running into is related to the "Calculate Statistics" step in the process of extracting mean spectra from an image qub. About nine times out of ten, after I have drawn an ROI and hit "Calculate Statistics," the ROI Statistics View window pops up with no data. This is particularly confusing, because when I extract a z-profile from any single pixel within the observation, there is clearly data present. Even more perplexing, is that the "Pixel Count" in the bottom of the ROI Statistics View window is labeled as zero.
Can anyone help me understand why this is happening? It is maddening, and it is impacting my ability to complete my research for my Master's degree.
Warm Regards,
-C
|
|
|
|
MariM
Veteran Member
Posts:2396 
| 03 Dec 2019 08:54 AM |
|
Can you try this in ENVI Classic? It could be an issue with the ROI dialog and the number of bands in the spectrometer - then trying to display the full statistics for each band. If you have access to IDL, you can also try http://www.harrisgeospati...IStatisticsTask.html |
|
|
|
Cory Hughes
New Member
Posts:6
| 03 Dec 2019 12:13 PM |
|
|
It does work in classic. It is good that I can get it to work, but the work around to classic is a tad time-intensive, because I need to port the stats back into the new version to do "curve smoothing" or boxcar averaging of the spectra.
Thanks for the suggestion.
I do have access to IDL, but I am very new at working with it, so that may be a more time-intensive process that just moving back and forth from Classic to the new version of ENVI.
Frustrating that it's not working in the new version. It has in the past. This issue just started happening recently, like in the last month.
|
|
|
|
MariM
Veteran Member
Posts:2396
| 03 Dec 2019 12:58 PM |
|
|
If you are saying that you used to be able to display the statistics for this file in ENVI previously, then something has changed.
You might try deleting your .idl directory from your home. For example, on windows this location is:
C:\Users\mminari\.idl
Then restart ENVI and try to calculate stats.
|
|
|
|
Cory Hughes
New Member
Posts:6
| 03 Dec 2019 01:31 PM |
|
|
Yes, this issue is recent, and periodically (maybe 1 out of 20 times I try the exact same set of steps) it doesn't occur and the statistics are calculated.
Deleting the .idl directory had no effect. :/
|
|
|
|
Cory Hughes
New Member
Posts:6
| 03 Dec 2019 02:20 PM |
|
|
Could this be a hardware issue? I have 16gb of ram, so I can't imagine that the calculation is crashing on account of maxing out the memory. However, the computer I am using does not have a great GPU and the CPU is middling.
|
|
|
|
Cory Hughes
New Member
Posts:6
| 03 Dec 2019 04:45 PM |
|
|
I've now tried this on a different computer, and I am getting similar issues, and still no usable results.
When I draw an ROI of ~50 pixels, the calculate statistics window is returning an average of either 5 pixels or zero pixels, but not the full ROI.
I can draw several ROIs on the same spot, and some of them will return a result calculating the statistics (though still erroneous results e.g., the sentence above), and others won't.
As stated, these issues persist between computers, but also between different CRISM observations. This leads me to believe that the root cause of the problem is not the computers I have access to, the data I am working with, but rather the issue is with ENVI.
Please help me solve this problem.
|
|
|
|
MariM
Veteran Member
Posts:2396
| 04 Dec 2019 05:33 AM |
|
|
If it is hardware, then it will be the graphics card. Try launching the IDLDE, then go to Windows->Preferences->IDL->Graphics. Change the rendering method from hardware to software. Then launch ENVI and try the stats.
|
|
|
|
Cory Hughes
New Member
Posts:6
| 04 Dec 2019 10:00 AM |
|
|
Thanks for the suggestion, but this did not fix the issue.
This is a fairly critical bug. Any chance Harris Geospatial could address this problem?
|
|
|
|
MariM
Veteran Member
Posts:2396
| 04 Dec 2019 01:50 PM |
|
|
You would need to come through Technical Support so that we can obtain the data that reproduces the problem. It is not an issue that has been reported before. If you are using a license from your institution, then submit it through the person who administers the license.
|
|
|
|
Cory Hughes
New Member
Posts:6
| 04 Dec 2019 06:17 PM |
|
|
Okay. I'll work with them on that. Thanks for all of your feedback and help. It's sincerely appreciated.
All the best,
-C
|
|
|
|