Home »
Public Forums »
archive »
IDL / ENVI help with MNF masking
IDL / ENVI help with MNF masking [message #13101] |
Sun, 11 October 1998 00:00 |
superman
Messages: 1 Registered: October 1998
|
Junior Member |
|
|
I am attempting to run an MNF transformation on an image that has been GPS
corrected causing alot of 'black' areas surrounding the image to the point
that the image is a narrow diaganol strip within the .img file. All of this
'black' (zero values) destroy the MNF image.
RSI tech support told me to "determine the mean spectrum of all the pixels
that contain meaningful data (those not zero), and then assign this spectrum
to all of the background pixels (zero values). --This is not a simple task in
ENVI since it must be accomplished for each individual band and could take me
a week per scene just to accomplis this.
RSI further said, "I realize this a really cumbersome process. If you have
experience with IDL and ENVI programming, you may want to write a procedure
that will run in batch mode or as a user function. That way you would not
need to repeat all of the steps interactively for each band. Plus you could
use the same code to do similar processing to other images i n the future. To
do the same sequence of events in an ENVI/IDL procedure you would probably
want to first make your mask in interactive ENVI or construct it using IDL
routines such as WHERE. You can use ENVI_STATS_DOIT to calculate the mean
spectrum using the mask. Then you would use ENVI_MASK_APPLY_DOIT to apply the
mask using the mean values, and CF_DOIT to create your new meta file with
different masked bands. Then you could use, MNF_DOIT to do the MNF transform"
Well isn't that easy! Not if you are not an ENVI or IDL programmer which I am
not. It would seem to me since MNF transforms are so critical to
hyperspectral processing ENVI would allow masks to be applied during the
transform. In the mean time I could use any help with this problem. I could
even send a dataset to demonstrate the problem. I just need to be able to
mask out all the zero values before doing an MNF transform.
Thanks for any help
Daniel Puchalski
(w) 407-494-7381
(f) 407-494-9439
(E-mail) hyperspectral@mindspring.com
|
|
|
Current Time: Wed Oct 08 15:54:28 PDT 2025
Total time taken to generate the page: 0.00408 seconds