Apply Forward MNF Transform to Spectra - NV5 Geospatial

Prev Go to previous topic

Go to next topic

Last Post 24 Apr 2012 06:02 PM by anon

Apply Forward MNF Transform to Spectra

1 Replies

Sort:


You are not authorized to post a reply.

Author

Messages

anon

New Member

Posts:

24 Apr 2012 06:02 PM
I applied a forward MNF transform to ASD spectra in order to create an MTMF classification endmember. The MNF scores range from -40 to 40 for the ASD transformed spectra while the MNF transformed HyMap image has scores that on average range from -2 to 2. (I first resampled the ASD spectra to HyMap bands before applying an MNF transform). I checked to make sure that the number of input bands used to calculate the statistics for the HyMap MNF transform was the same as the number of output bands. I can run a classification, but the results tend to tease out targets with bright reflectance rather than sagebrush. Also, I do not suspect that it is the ASD data itself causing this problem. Any thoughts? Thanks, Jessica Mitchell

MariM

Veteran Member

Posts:2396
Veteran Member

25 Apr 2012 08:15 AM
I think the first thing I would look at is if the data cube and spectra are in the same data space - i.e. are both in reflectance between 0-1.0? I assume the spectra taken from the ASD were converted to reflectance but what about the Hymap data? If the Hymap data was converted to reflectance, do the values range between 0-1.0 or are they scaled to integers, which is a common practice in many atmospheric correction programs such as FLAASH? I would check these things first to make sure the datasets are comparable.

You are not authorized to post a reply.