Blind Image Fusion for Hyperspectral Imaging with the Directional Total Variation

  • 2017-12-14 13:50:06
  • Leon Bungert, David A. Coomes, Matthias J. Ehrhardt, Jennifer Rasch, Rafael Reisenhofer, Carola-Bibiane Schönlieb
  • 0

Abstract

Hyperspectral imaging is a cutting-edge type of remote sensing used formapping vegetation properties, rock minerals and other materials. A majordrawback of hyperspectral imaging devices is their intrinsic low spatialresolution. In this paper, we propose a method for increasing the spatialresolution of a hyperspectral image by fusing it with an image of higherspatial resolution that was obtained with a different imaging modality. This isaccomplished by solving a variational problem in which the regularizationfunctional is the directional total variation. To accommodate for possiblemis-registrations between the two images, we consider a non-convex blindsuper-resolution problem where both a fused image and the correspondingconvolution kernel are estimated. Using this approach, our model can realignthe given images if needed. Our experimental results indicate that thenon-convexity is negligible in practice and that reliable solutions can becomputed using a variety of different optimization algorithms. Numericalresults on real remote sensing data from plant sciences and urban monitoringshow the potential of the proposed method and suggests that it is robust withrespect to the regularization parameters, mis-registration and the shape of thekernel.

 

Quick Read (beta)

loading the full paper ...