Dictionary Learning-Cooperated Matrix Decomposition for Hyperspectral Target Detection

Hyperspectral target detection is one of the most challenging tasks in remote sensing due to limited spectral information. Many algorithms based on matrix decomposition (MD) are proposed promote separation background and targets, but they suffer from two problems: (1) Targets detected with criterion reconstruction residuals, imbalanced number atoms union dictionary may lead misclassification targets. (2) The results susceptible quality apriori spectra, thus obtaining inferior performance because inevitable variability. In this paper, we propose a decomposition-based detector named learning-cooperated (DLcMD) for hyperspectral detection. procedure DLcMD two-fold. First, low rank sparse (LRaSMD) exploited separate targets its insensitivity atoms, which can reduce Inspired by learning, updated during LRaSMD alleviate impact After that, binary hypothesis model specifically designed proposed, generalized likelihood ratio test (GLRT) performed obtain final result. Experimental five datasets have shown reliability method. Especially Los Angeles-II dataset, area under curve (AUC) value nearly 16% higher than average other seven detectors, reveals superiority