Matched subspace detector based on sparse representation for target detection in hyperspectral imagery

Abstract

In this paper, we propose a new sparsity-based matched subspace detector (SMSD) algorithm for target detection in hyperspectral imagery (HSI). This algorithm, based on the concept that a pixel in HSI lies in a low-dimensional subspace and thus can be represented as a sparse linear combination of the training samples, explores the linear mixing model to both specify the desired target and characterizes the interfering background. In the algorithm, the linear subspace mixture model for the MSD is first reformulated and then the corresponding expression for the generalized likelihood ratio test (GLRT) is obtained for this model and at last we use the sparse representation to modify the MSD model. The proposed algorithm is compared with the sparsity-based algorithm. Simulation results show that our algorithm outperforms the sparsity-based target detection algorithm.