A New Elbow Estimation Method for Selecting the Best Solution in Sparse Unmixing

The goal of hyperspectral image analysis is often to determine which materials, out a given set possibilities, are present in each pixel. As data being gathered rapidly increasing amounts, automatic becoming progressively more important. Automatic identification materials from mixed pixel possible with 1) Bayesian unmixing algorithms and 2) multiobjective sparse when method such as elbow estimation used select the best solution Pareto-optimal solutions. We develop new called termination condition adaptive (TCAE) for selecting solutions biobjective problem. Specifically, two objectives assumed be sparsity level fractional abundance vector reconstruction error. conduct experiments real-world applications mind, TCAE performs significantly better than state-of-the-art they both sequence vectors generated by iterative spectral mixture (ISMA). Furthermore, combination ISMA able identify endmembers pixels several times faster higher F1-score reference. conclude that facilitates automatic, reliable, rapid pixels.