Deep learning reconstruction for single pixel imaging with generative adversarial networks

Simple item page
buir.contributor.authorGüven, Baturalp
buir.contributor.authorÇukur, Tolga
buir.contributor.orcidÇukur, Tolga|0000-0002-2296-851X
dc.citation.epage2064en_US
dc.citation.spage2060
dc.contributor.authorGüven, Baturalp
dc.contributor.authorGüngör, A.
dc.contributor.authorBahçeci, M. U.
dc.contributor.authorÇukur, Tolga
dc.coverage.spatialKuala Lumpur, Malaysia
dc.date.accessioned2024-03-07T10:38:14Z
dc.date.available2024-03-07T10:38:14Z
dc.date.issued2023-09-11
dc.departmentDepartment of Electrical and Electronics Engineering
dc.descriptionConference Name: 2023 IEEE International Conference on Image Processing (ICIP)
dc.descriptionDate of Conference: 08-11 October 2023
dc.description.abstractSingle pixel imaging (SPI) enables high-resolution imaging through multiple coded measurements based on low-resolution snapshots. An inverse problem can then be solved to reconstruct a high-resolution image given the coded measurements. There has been recent interest in adoption of deep neural networks in SPI reconstruction. However, existing methods are commonly trained with pixel-wise loss terms such as the ℓ 1 -norm loss, which can result in spatial blurring and poor sensitivity to structural details. In this study, we propose a novel approach for deep SPI reconstruction based on an unrolled conditional generative adversarial network (cGAN) model. The generator estimates the high-resolution image using coded low-resolution measurements by iterating across a cascade of denoising and data-consistency modules. Meanwhile, the discriminator distinguishes real versus synthesized high-resolution images. The architecture is trained end-to-end via a combined pixel-wise and adversarial loss to enhance sensitivity to structural details. The proposed method is demonstrated against existing SPI reconstruction methods, and ablation studies are performed to demonstrate the individual model components. The proposed method outperforms competing methods in terms of both quantitative metrics and visual quality.
dc.identifier.doi10.1109/ICIP49359.2023.10223149
dc.identifier.eisbn978-1-7281-9835-4
dc.identifier.isbn978-1-7281-9836-1
dc.identifier.urihttps://hdl.handle.net/11693/114387
dc.language.isoen_US
dc.publisherIEEE
dc.relation.isversionofhttps://dx.doi.org/10.1109/ICIP49359.2023.10223149
dc.source.title2023 IEEE International Conference on Image Processing (ICIP)
dc.subjectConditional generative adversarial network (cGAN)
dc.subjectSingle pixel imaging (SPI)
dc.subjectAdversarial loss
dc.subjectPlug and play
dc.subjectℓ1 Loss
dc.titleDeep learning reconstruction for single pixel imaging with generative adversarial networks
dc.typeConference Paper

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Deep_learning_reconstruction_for_single_pixel_imaging_with_generative_adversarial_networks.pdf
Size:
3.97 MB
Format:
Adobe Portable Document Format
Download
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
2.01 KB
Format:
Item-specific license agreed upon to submission
Description:
Download

Collections

Department of Electrical and Electronics Engineering