@inproceedings{7cf756c3d16742b2b3f806089cff2458,
title = "Partial unmixing of hyperspectral imagery: Theory and methods",
abstract = "The Minimum Noise Fraction (MNF) data reduction transform and Mixture Tuned Matched Filtering (MTMF) partial unmixing classification algorithm are relatively new image processing techniques that have proven to be effective target detection tools. These techniques allow partial unmixing and subpixel target abundance estimation, products that cannot be simultaneously achieved using standard mixture modeling or spectral angle mapping algorithms. This paper presents a tangible description of the technical background of both algorithms, a resource that is currently unavailable in existing literature. A demonstration of the use of the MNF and MTMF techniques is presented in detail for application to leafy spurge infestations in the Swan Valley, Idaho. The use of these techniques on hyperspectral imagery generated a producer's accuracy of 63% for infestations with canopy cover averaging 40% for imagery with 3.5 m resolution. Ramifications of image noise estimation and classification endmember selection are discussed at length and should be used as a resource guide for application to other vegetation studies.",
author = "Mundt, {Jacob T.} and Streutker, {David R.} and Glenn, {Nancy F.}",
year = "2007",
language = "English",
isbn = "9781604232240",
series = "American Society for Photogrammetry and Remote Sensing - ASPRS Annual Conference 2007: Identifying Geospatial Solutions",
pages = "440--451",
booktitle = "American Society for Photogrammetry and Remote Sensing - ASPRS Annual Conference 2007",
note = "ASPRS Annual Conference 2007: Identifying Geospatial Solutions ; Conference date: 07-05-2007 Through 11-05-2007",
}