Should spectralDist, abundanceImage, cos.dist be part of unmixR?

[rguliev]

In GitLab by @cbeleites on Oct 6, 2015, 16:30

Created by: cbeleites

CB says: in my opinion, they are not really related to spectral unmixing:

• spectralDist (and cos.dist?) may be best in ChemoSpec?
• I could also imagine hosting cos.dist in hyperSpec (which already has pearson.dist)
• abundanceImage: do we need that (e.g. as hyperSpec::plotmap is awfully slow for raster images)?
  • I could imagine adding some functionality that extracts raster image data cubes from hyperSpec objects.
  • or we just source abundanceImage for unmixR's vignettes etc.

Imported comments:

By bryanhanson on 2014-06-12 17:48:11 UTC

Well, I wrote spectralDist so that the endmembers could be easily compared to authentic spectra. I was doing that to test if we were getting the right answers. If you want to keep it, unmixR will need amap as a dependency, and package lsa has a function called cosine that could be grabbed or used as a dependency since I got stalled on writing a correct cosineAlpha helper function.

I wrote abundanceImage just to make sure that I understood how this should work (and didn’t realize that plotmap existed).

So I see both of these as conveniences. If you feel they are too far from the heart of unmixR I’ll move them to ChemoSpec.

It’s your call.

Bryan

On Jun 12, 2014, at 1:34 PM, cbeleites notifications@github.com wrote:

CB says: in my opinion, they are not really related to spectral unmixing:

spectralDist (and cos.dist?) may be best in ChemoSpec? I could also imagine hosting cos.dist in hyperSpec (which already has pearson.dist) abundanceImage: do we need that (e.g. as hyperSpec::plotmap is awfully slow for raster images)? I could imagine adding some functionality that extracts raster image data cubes from hyperSpec objects. or we just source abundanceImage for unmixR's vignettes etc. — Reply to this email directly or view it on GitHub.

By cbeleites on 2014-06-12 17:59:02 UTC

• I wrote a cos.dist today (which I believe is working as expected :) ), so we don't need yet another dependency.
• if we need the comparison in the vignettes, we can just use vignetteDepends and Suggests: amap (if we need any of their distances).
• let's drop abundanceImage in favor of hyperSpec::plotmap - we anyways need hyperSpec for the ENVI import.

[rguliev]

In GitLab by @bryanhanson on Jun 2, 2016, 19:16

Re-opening this issue as we never removed the files!

[rguliev]

In GitLab by @bryanhanson on Jun 14, 2016, 20:23

As part of the namespace clean up I am doing on branch VCALopezBugFix I went head and removed these files.