A Low-rank Tensor Regularization Strategy for Hyperspectral Unmixing

摘要: Tensor-based methods have recently emerged as a more natural and effective formulation to address many problems in hyperspectral imaging. In unmixing (HU), low-rank constraints on the abundance maps been shown act regularization which adequately accounts for multidimensional structure of underlying signal. However, imposing strict constraint does not seem be adequate, important information that may required represent fine scale behavior discarded. This paper introduces new tensor captures without hindering flexibility solution. Simulation results with synthetic real data show extra introduced by proposed significantly improves results.