Linear and nonlinear unmixing in hyperspectral imaging

摘要: Abstract Mainly due to the limited spatial resolution of data acquisition devices, hyperspectral image pixels generally result from mixture several components that are present in observed surface. Spectral analysis (or spectral unmixing) is a key processing step which aims at identifying signatures these materials and quantifying their distribution over image. The main purpose this chapter introduce unmixing problem discuss some linear nonlinear models algorithms used solve it. We will show that, capitalizing on decades methodological developments geoscience remote sensing community, most proposed unmix remotely sensed images can be directly applied chemometrics field process arising various scanning microscopic techniques such as transmission electron microscopy Raman imaging.