GEOSCAN Search Results: Fastlink

GEOSCAN Menu


TitleAutomatic Endmember Selection: Part 1 Theory
 
AuthorSzeredi, T; Staenz, K; Neville, R A
SourceRemote Sensing of Environment 1999.
Year1999
Alt SeriesEarth Sciences Sector, Contribution Series 20042703
PublisherElsevier
Documentserial
Lang.English
Mediapaper
AbstractThis paper presents four algorithms for automatic extraction of endmembers from imaging spectrometer data. Two of the algorithms, the Iterative Target Transform Factor Analysis (ITTFA) and the Alternating Regression (AR) methods do not rely on pure pixels and can be used in cases where every pixel in the scene is mixed. The Iterative Error Analysis (IEA) method makes use of constrained unmixing to successively choose endmembers which eliminate the errors in the unmixing. The Purest Pixel Clustering (PPC) method uses the pixel purity index to assign weights to the purest pixels and then creates endmembers by performing a weighted clustering of these pure pixels. The IEA and PPC methods are most useful in extracting endmembers from scenes which have relatively pure pixels.
GEOSCAN ID219505

 
Date modified: