eduzhai > Applied Sciences > Engineering >

Coupled Convolutional Neural Network with Adaptive Response Function Learning for Unsupervised Hyperspectral Super-Resolution

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 16 pages

Abstract: Due to the limitations of hyperspectral imaging systems, hyperspectralimagery (HSI) often suffers from poor spatial resolution, thus hampering manyapplications of the imagery. Hyperspectral super-resolution refers to fusingHSI and MSI to generate an image with both high spatial and high spectralresolutions. Recently, several new methods have been proposed to solve thisfusion problem, and most of these methods assume that the prior information ofthe Point Spread Function (PSF) and Spectral Response Function (SRF) are known.However, in practice, this information is often limited or unavailable. In thiswork, an unsupervised deep learning-based fusion method - HyCoNet - that cansolve the problems in HSI-MSI fusion without the prior PSF and SRF informationis proposed. HyCoNet consists of three coupled autoencoder nets in which theHSI and MSI are unmixed into endmembers and abundances based on the linearunmixing model. Two special convolutional layers are designed to act as abridge that coordinates with the three autoencoder nets, and the PSF and SRFparameters are learned adaptively in the two convolution layers during thetraining process. Furthermore, driven by the joint loss function, the proposedmethod is straightforward and easily implemented in an end-to-end trainingmanner. The experiments performed in the study demonstrate that the proposedmethod performs well and produces robust results for different datasets andarbitrary PSFs and SRFs.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×