FUTURE HYPERSPECTRAL IMAGING PLATFORMS

Spectral libraries are required for successful classification because each pure material has its own unique spectral signature which are essential for the classification of hyperspectral imagery. Each pixel within a hyperspectral image naturally contains more than one material; thus, determining the spectral mixture of pixels, also known as the spectral unmixing problem, is critical. Moreover, spectral signatures vary according to species, so that it is critical for Turkey to develop its own spectral library for indigenous plant varieties. It is estimated that there are over 200 species indigenous to Anatolia. Without a geography specific library, the results of hyperspectral image classification and spectral unmixing will be erroneous - i.e., results obtained with use of non-native libraries such as that of the USGS can yield flawed results.