This work presents a method for hyperspectral image unmixing based on nonnegative tensor factorization. While traditional approaches may process spectral information without regard for spatial structures in the data set, ...