Hyperspectral Image Restoration via Global <i>L_1-2 </i> Spatial–Spectral Total Variation Regularized Local Low-Rank Tensor Recovery

Hyperspectral images (HSIs) are usually corrupted by various noises, e.g., Gaussian noise, impulse stripes, dead lines, and many others. In this article, motivated the good performance of L 1-2 nonconvex metric in image sparse structure exploitation, we first develop a 3-D spatial-spectral total variation ( SSTV) regularization to globally represent prior gradient domain HSIs. Then, divide HSIs into local overlapping patches, low-rank tensor recovery (LTR) is locally used effectively separate clean HSI patches from complex noise. The patchwise LTR can not only adapt property well but also significantly reduce information loss caused global LTR. Finally, integrating advantages both SSTV model, propose regularized model for hyperspectral restoration. framework alternating direction method multipliers, difference convex algorithm, split Bregman iteration method, singular value decomposition adopted solve proposed efficiently. Simulated real experiments show that dependence on noise independent identical distribution hypotheses, simultaneously remove types even structure-related