|
Hi Tyas,
1. Yes, you will probably get similar results after you switch to unconstrained unmixing, because your endmembers are probably not adequate to model your scene. But at least with unconstrained unmixing you will know for sure that you have a good set of endmembers when those problems largely go away. With constrained unmixing you never know whether good looking results are due to the contraining, or to the fact that you really have a good model for the scene.
So, in summary, you need to take a good look at your endmember spectra, and try to figure out what is wrong. Maybe you are missing one or more spectra for materials that are in the scene? Remember that with unmixing, you need to have spectra for all of the materials in the scene, or else the model will not work. Or maybe you are using spectra that are not very representative of pure forms of some of your materials?
3. Did you try mixture tuned match filtering? That gives you an infeasibility band, in addition to the match score that regular match filtering provides. The infeasibility helps you see where you have false positives (which match filtering is prone to). Maybe that will help? Also, keep in mind that match filtering is only appropriate if the target you are looking for is rare within the scene. If it is very common, then the assumption that the overall image stats are representative of the background to the target is not valid.
4. I'm not sure what you mean when you say that in your observations the MNF have minimum values. If you can provide more details about what you mean, I might be able to comment.
The algorithm used for ENVI's MNF transform is based on those found in the following references:
Green, A. A., Berman, M., Switzer, P., and Craig, M. D., 1988, A transformation for ordering multispectral data in terms of image quality with implications for noise removal: IEEE Transactions on Geoscience and Remote Sensing, v. 26, no. 1, p. 65-74.
Boardman, J. W., and Kruse, F. A., 1994, Automated spectral analysis: a geological example using AVIRIS data, north Grapevine Mountains, Nevada: in Proceedings, ERIM Tenth Thematic Conference on Geologic Remote Sensing, Environmental Research Institute of Michigan, Ann Arbor, MI, pp. I-407 - I-418.
ENVI's Linear Spectral Unmixing algorithm is based on those found in the following references:
Boardman, J. W., 1989, Inversion of imaging spectrometry data using singular value decomposition: in Proceedings, IGARSS'89, 12th Canadian Symposium on Remote Sensing, v. 4., pp. 2069-2072.
Boardman, J. W., 1992, Sedimentary facies analysis using imaging spectrometry: A geophysical inverse problem: Unpublished Ph. D. Thesis, University of Colorado, Boulder, p. 212.
I hope this helps.
Peg
|