| 01 Dec 2014 10:08 AM |
|
|
Hi all,
I am working on ASTER L1B imagery. Thermal bands (ASTER TIR) were atmospherically corrected and thermal emissivity bands were calculated. I am trying to reduce the data noise using MNF - Inverse MNF transformation method, but every time I tried to do so, the process stopped at 99% and not responding at all without giving any error messages. I am using ENVI 5.1 program.
This is the statistics of the five emissivity bands:
Basic Stats Min Max Mean Stdev Num Eigenvalue
Band 1 0.858331 0.960000 0.909946 0.011863 1 0.000000
Band 2 0.857928 0.953124 0.916422 0.011732 2 0.000000
Band 3 0.854245 0.960000 0.924670 0.012574 3 0.000000
Band 4 0.941393 0.960000 0.958717 0.002011 4 0.000000
Band 5 0.939066 0.960000 0.958404 0.002219 5 0.000000
Thanks for your help
Mahmoud
|
|
|
|
MariM
Veteran Member
Posts:2396  
| 02 Dec 2014 10:06 AM |
|
|
It is hard to say. MNF requires a matrix inversion to calculate covariance and I suspect your emissivity bands have very little dynamic range so it may be unable to complete the inversion. I would try the process in ENVI Classic. Launch ENVI+IDL and at the IDL command prompt type:
ENVI>envi_toggle_catch
Then try the process again. If it stops, it may be a crash and output some information to the IDL console that would provide a better clue as to what is going on.
|
|
|
|
Stavroula Giannakopoulou
New Member
Posts:49 
| 21 Nov 2019 02:30 PM |
|
|
Hello everyone,
I have a question regarding to the procedure of MNF. The thing is that I implemented at my Hyperion data the MNF transformation, and I also select a specific number of bands. My question is whether I can classify the result that I receive, through the SAM method, and the usage of a number of spectral signatures from a sectral library. Thank you very much for your help!
|
|
|
|
MariM
Veteran Member
Posts:2396
| 22 Nov 2019 10:04 AM |
|
|
You could run classification on your MNF images with the spectral signatures extracted from the image but they must be in the same spectral space. So you would need to perform an MNF transform on the spectral library using "Apply forward MNFs to spectra' using the MNF statistics calculated on the image. Then they can be used together.
|
|
|
|
Stavroula Giannakopoulou
New Member
Posts:49
| 22 Nov 2019 11:29 AM |
|
|
Thank you very much for your quick response! Despite the fact that I have done the resampling at the spectral library of interest, when I tried to implement for example the SAM method, it shows me the error that: "the input spectrum does not contain the same number of bands, as the target and no resampling can be done without wavelengths"... Any idea about this. Something wrong with the resampling?
|
|
|
|
MariM
Veteran Member
Posts:2396
| 22 Nov 2019 03:23 PM |
|
|
Did you mention you reduced the number of MNF bands? You are probably running up against this issue: https://www.harrisgeospatial.com/Support/Self-Help-Tools/Help-Articles/Help-Articles-Detail/ArtMID/10220/ArticleID/19017/3876
|
|
|
|
Stavroula Giannakopoulou
New Member
Posts:49
| 22 Nov 2019 03:48 PM |
|
|
Yes, of course, I subsetted the MNF image, so as to have the first 15 bands. I will check this article! I think it will solve the problem! Thank you very much for your help!
|
|
|
|
Stavroula Giannakopoulou
New Member
Posts:49
| 23 Nov 2019 12:19 PM |
|
|
That was very helpful! I will try to execute the procedure on IDL! Is there any tutorial for beginners, so as to use it? Thank you very much for your help!
|
|
|
|
MariM
Veteran Member
Posts:2396
| 25 Nov 2019 05:17 AM |
|
|
I don't think there is any specific tutorial for the steps in the help article. However, if you open the spectral library in ENVI Classic, you can use the Subset tool to subset the library, as all spectral libraries are stored as image files (.dat+.hdr) with each material as a line and each band as the sample. This is discussed at the bottom of the article.
|
|
|
|
Stavroula Giannakopoulou
New Member
Posts:49
| 01 Dec 2019 05:31 AM |
|
|
I tried to implement the instructions from the specific tutorial. But now the problem is that I think that at the previous step I think that I should firstly implement the Forward MNF to spectra, for the spectral signatures that I have from the spectral libraries... The problem is that each material's signature has the same plot at the diagram, despite the fact that at the libraries it doesn't and I don't know why. Any help?
I
|
|
|
|