LEAST-SQUARES SUBSPACE PROJECTION APPROACH TO MIXED PIXEL CLASSIFICATION FOR HYPERSPECTRAL IMAGES

An orthogonal subspace projection (OSP) method using linear mixture mo deling was recently explored in hyperspectral image classification and has shown promise in signature detection, discrimination, and classif ication. In this paper, the OSP is revisited and extended by three unc onstrained least squares subspace protection approaches, called signat ure space OSP, target signature space OSP, and oblique subspace projec tion, where the abundances of spectral signatures are not known a prio ri hut need to be estimated, a situation to which the OSP cannot be di rectly applied, The proposed three subspace projection methods can be used not only to estimate signature abundance, but also to classify a target signature at subpixel scale so as to achieve subpixel detection . As a result, they can be viewed as a posteriori OSP as opposed to OS P, which can be thought of as a priori OSP, In order to evaluate these three approaches, their associated least squares estimation errors ar e cast as a signal detection problem in the framework of the Neyman-Pe arson detection theory so that the effectiveness of their generated cl assifiers can be measured by characteristics (ROC) analysis, All resul ts are demonstrated by computer simulations and Airborne Visible/Infra red Imaging Spectrometer (AVIRIS) data.