Fast and High-Quality Blind Multi-Spectral Image Pansharpening

Blind pansharpening addresses the problem of generating a high spatial-resolution multi-spectral (HRMS) image given low (LRMS) with guidance its associated spatially misaligned panchromatic (PAN) without parametric side information. In this paper, we propose fast approach to blind and achieve state-of-the-art reconstruction quality. Typical algorithms are often computationally intensive since blur kernel target HRMS computed using iterative solvers in an alternating fashion. To pansharpening, decouple solution image. First, estimate by computing coefficients minimum total generalized variation that downsampled version PAN approximate linear combination LRMS channels. Then, each channel local Laplacian prior regularize relationship between Solving is accelerated both parallelizing across channels numerical for channel. Due scheme powerful priors used on (total variation) cross-channel (local prior), experiments demonstrate our algorithm outperforms model-based counterparts terms computational time quality images.