Hyperspectral Image Deconvolution with a Spectral-Spatial Total Variation Regularization

Abstract

Hyperspectral images are often unavoidably degraded by blur and noise in the acquisition process, which influences subsequent image processing and analysis. Under the maximum a posteriori framework, this article introduces a hyperspectral image deconvolution method based on spectral-spatial total variation prior and an explicit nonnegative constraint, which can preserve the edge in the spatial domain and maintain the discontinuity along the spectral dimension. Additionally, the proposed method fully exploits the spectral correlation between adjacent bands. The alternating direction method of multipliers (ADMM) is employed to efficiently optimize the proposed spectral-spatial total variation deconvolution model. The reason for this interest is that ADMM decomposes a difficult optimization problem into a sequence of much simpler and easily solved sub-problems. The regularization parameter is updated adaptively to make the proposed algorithm more stable. Numerical experiments are conducted in both simulated and real data to verify the performance of the proposed method.