II
Endmember Extraction
Due to significantly improved high spatial and spectral resolution provided by hyperspectral imaging sensors endmember extraction has become increasingly important in hyperspectral image analysis. According to the definition given in Schwengerdt (1997), an endmember is an idealized, pure signature for a class, more specifically, spectral class. For multispectral imagery, an endmember may be difficult to find, due to the fact that many image pixels are generally heavily mixed because of its low spatial and spectral resolution. As a result, endmember extraction has received little interest in the past and its importance has been overlooked. By contrast, with recent advances in hyperspectral imaging sensors many subtle material substances that cannot be resolved by multispectral imagery can now be uncovered by hyperspectral imagery. These substances are generally not known a priori and can be only diagnosed by high spectral resolution. Endmembers are considered to be one of such substances. Their existence in image data cannot be generally detected by human eye. Most importantly, once endmembers are present, they may be very likely to appear as anomalies and their sample pools may be relatively small. Because of such characteristics, finding endmembers is very challenging. Many algorithms have been developed for this purpose, such as pixel purity index (PPI) (Boardman et al., 1995), N-finder algorithm (N-FINDR) (Winter, 1999a,b), iterative error analysis (IEA) ...
