Deep Joint Demosaicking and Super-Resolution for Spectral Filter Array Images

In spectral imaging, the constraints imposed by hardware often lead to a limited spatial resolution within spectral filter array images. On the other hand, the process of demosaicking is challenging due to intricate filter patterns and a strong spectral cross correlation. Moreover, demosaicking and...

Full description

Saved in:
Bibliographic Details
Main Authors: Abdelhamid N. Fsian, Jean-Baptiste Thomas, Jon Y. Hardeberg, Pierre Gouton
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10838565/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1832583240793194496
author Abdelhamid N. Fsian
Jean-Baptiste Thomas
Jon Y. Hardeberg
Pierre Gouton
author_facet Abdelhamid N. Fsian
Jean-Baptiste Thomas
Jon Y. Hardeberg
Pierre Gouton
author_sort Abdelhamid N. Fsian
collection DOAJ
description In spectral imaging, the constraints imposed by hardware often lead to a limited spatial resolution within spectral filter array images. On the other hand, the process of demosaicking is challenging due to intricate filter patterns and a strong spectral cross correlation. Moreover, demosaicking and super resolution are usually approached independently, overlooking the potential advantages of a joint solution. To this end, we use a two-branch framework, namely a pseudo-panchromatic image network and a pre-demosaicking sub-branch coupled with a novel deep residual demosaicking and super resolution module. This holistic approach ensures a more coherent and optimized restoration process, mitigating the risk of error accumulation and preserving image quality throughout the reconstruction pipeline. Our experimental results underscore the efficacy of the proposed network, showcasing an improvement of performance both qualitatively and quantitatively when compared to the sequential combination of state-of-the-art demosaicking and super resolution. With our proposed method, we obtained on the ARAD-1K dataset an average PSNR of 48.02 (dB) for domosaicking only, equivalent to the best method of the state-of-the-art. Moreover, for joint demosaicking and super resolution our model averages 35.26 (dB) and 26.29 (dB), respectively for <inline-formula> <tex-math notation="LaTeX">$\times 2$ </tex-math></inline-formula> and <inline-formula> <tex-math notation="LaTeX">$\times 4$ </tex-math></inline-formula> upscale, outperforming state-of-the-art sequential approach.The codes and datasets are available at <uri>https://github.com/HamidFsian/DRDmSR</uri>.
format Article
id doaj-art-8e5418745ecb4b6c99906c1b2e130166
institution Kabale University
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-8e5418745ecb4b6c99906c1b2e1301662025-01-29T00:01:19ZengIEEEIEEE Access2169-35362025-01-0113162081622210.1109/ACCESS.2025.352875310838565Deep Joint Demosaicking and Super-Resolution for Spectral Filter Array ImagesAbdelhamid N. Fsian0https://orcid.org/0009-0006-5388-7428Jean-Baptiste Thomas1Jon Y. Hardeberg2https://orcid.org/0000-0003-1150-2498Pierre Gouton3Department Informatique, Electronique, M&#x00E9;canique (IEM), Imagerie et Vision Artificielle (ImVIA) Laboratory, Universit&#x00E9; de Bourgogne, Dijon, FranceDepartment Informatique, Electronique, M&#x00E9;canique (IEM), Imagerie et Vision Artificielle (ImVIA) Laboratory, Universit&#x00E9; de Bourgogne, Dijon, FranceColourlab, Department of Computer Science, Norwegian University of Science and Technology (NTNU), Gj&#x00F8;vik, NorwayDepartment Informatique, Electronique, M&#x00E9;canique (IEM), Imagerie et Vision Artificielle (ImVIA) Laboratory, Universit&#x00E9; de Bourgogne, Dijon, FranceIn spectral imaging, the constraints imposed by hardware often lead to a limited spatial resolution within spectral filter array images. On the other hand, the process of demosaicking is challenging due to intricate filter patterns and a strong spectral cross correlation. Moreover, demosaicking and super resolution are usually approached independently, overlooking the potential advantages of a joint solution. To this end, we use a two-branch framework, namely a pseudo-panchromatic image network and a pre-demosaicking sub-branch coupled with a novel deep residual demosaicking and super resolution module. This holistic approach ensures a more coherent and optimized restoration process, mitigating the risk of error accumulation and preserving image quality throughout the reconstruction pipeline. Our experimental results underscore the efficacy of the proposed network, showcasing an improvement of performance both qualitatively and quantitatively when compared to the sequential combination of state-of-the-art demosaicking and super resolution. With our proposed method, we obtained on the ARAD-1K dataset an average PSNR of 48.02 (dB) for domosaicking only, equivalent to the best method of the state-of-the-art. Moreover, for joint demosaicking and super resolution our model averages 35.26 (dB) and 26.29 (dB), respectively for <inline-formula> <tex-math notation="LaTeX">$\times 2$ </tex-math></inline-formula> and <inline-formula> <tex-math notation="LaTeX">$\times 4$ </tex-math></inline-formula> upscale, outperforming state-of-the-art sequential approach.The codes and datasets are available at <uri>https://github.com/HamidFsian/DRDmSR</uri>.https://ieeexplore.ieee.org/document/10838565/Spectral imagingdemosaickingsuper resolutiondeep learningpseudo-panchromatic imagespectral filter array
spellingShingle Abdelhamid N. Fsian
Jean-Baptiste Thomas
Jon Y. Hardeberg
Pierre Gouton
Deep Joint Demosaicking and Super-Resolution for Spectral Filter Array Images
IEEE Access
Spectral imaging
demosaicking
super resolution
deep learning
pseudo-panchromatic image
spectral filter array
title Deep Joint Demosaicking and Super-Resolution for Spectral Filter Array Images
title_full Deep Joint Demosaicking and Super-Resolution for Spectral Filter Array Images
title_fullStr Deep Joint Demosaicking and Super-Resolution for Spectral Filter Array Images
title_full_unstemmed Deep Joint Demosaicking and Super-Resolution for Spectral Filter Array Images
title_short Deep Joint Demosaicking and Super-Resolution for Spectral Filter Array Images
title_sort deep joint demosaicking and super resolution for spectral filter array images
topic Spectral imaging
demosaicking
super resolution
deep learning
pseudo-panchromatic image
spectral filter array
url https://ieeexplore.ieee.org/document/10838565/
work_keys_str_mv AT abdelhamidnfsian deepjointdemosaickingandsuperresolutionforspectralfilterarrayimages
AT jeanbaptistethomas deepjointdemosaickingandsuperresolutionforspectralfilterarrayimages
AT jonyhardeberg deepjointdemosaickingandsuperresolutionforspectralfilterarrayimages
AT pierregouton deepjointdemosaickingandsuperresolutionforspectralfilterarrayimages