Spatial subspace clustering for hyperspectral data segmentation

![CDATA[We propose a novel method called spatial subspace clustering (SpatSC) for 1D hyperspectral data segmentation problem, e.g. hyperspectral data taken from a drill hole. Addressing this problem has several practical uses such as improving interpretability of the data, and especially a better understanding of the mineralogy. Spatial subspace clustering is a combination of subspace learning and the fused lasso. As a result, it is able to produce spatially smooth clusters. From this point of view, it can be simply interpreted as a spatial information guided subspace learning algorithm. SpatSC has flexible structures that embrace the cases with and without library of pure spectra. It can be further extended, for example, using different error structures, such as including rank operator. We test this method on a real drill hole thermal infrared hyperspectral data set called DDH9. SpatSC produces stable and continuous segments, which are more interpretable. This property is not shared by other state-of-the-art subspace learning algorithms.]]