Single Pixel Imaging Based on Multiple Prior Deep Unfolding Network

Single-pixel imaging (SPI), an imaging technique based on the theory of compressed sensing, is limited in real-time imaging and high-resolution images due to its relatively slow imaging speed. In recent years, deep unfolding network compressed sensing reconstruction algorithms based on deep learning...

Full description

Saved in:
Bibliographic Details
Main Authors: Quan Zou, Qiurong Yan, Qianling Dai, Ao Wang, Bo Yang, Yi Li, Jinwei Yan
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Photonics Journal
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10577085/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849415946427432960
author Quan Zou
Qiurong Yan
Qianling Dai
Ao Wang
Bo Yang
Yi Li
Jinwei Yan
author_facet Quan Zou
Qiurong Yan
Qianling Dai
Ao Wang
Bo Yang
Yi Li
Jinwei Yan
author_sort Quan Zou
collection DOAJ
description Single-pixel imaging (SPI), an imaging technique based on the theory of compressed sensing, is limited in real-time imaging and high-resolution images due to its relatively slow imaging speed. In recent years, deep unfolding network compressed sensing reconstruction algorithms based on deep learning have proven to be an effective solution for faster and higher quality image reconstruction. However, existing deep unfolding networks mainly rely on a single piece of a priori information and may ignore other intrinsic structures of the image. Therefore, in this paper, we propose a deep unfolding network (MPDU-Net) that incorporates multiple prior information. To effectively fuse multiple prior information, we propose three different fusion strategies in the deep reconstruction sub-network. An unbiased convolutional layer is used to simulate the sampling reconstruction process to achieve joint reconstruction for effective removal of block artifacts. The sampling matrix is input into the deep reconstruction sub-network as a learnable parameter to achieve joint optimization of sampling reconstruction. Simulation and practical experimental results show that the proposed network outperforms existing compressed sensing reconstruction algorithms based on deep unfolding networks.
format Article
id doaj-art-a01365a4697c469ab2b700f77db16fe9
institution Kabale University
issn 1943-0655
language English
publishDate 2024-01-01
publisher IEEE
record_format Article
series IEEE Photonics Journal
spelling doaj-art-a01365a4697c469ab2b700f77db16fe92025-08-20T03:33:21ZengIEEEIEEE Photonics Journal1943-06552024-01-0116411010.1109/JPHOT.2024.342078710577085Single Pixel Imaging Based on Multiple Prior Deep Unfolding NetworkQuan Zou0https://orcid.org/0009-0003-7280-3953Qiurong Yan1https://orcid.org/0000-0003-4736-7435Qianling Dai2https://orcid.org/0009-0001-1735-2606Ao Wang3Bo Yang4https://orcid.org/0009-0002-0752-482XYi Li5Jinwei Yan6School of Information Engineering, Nanchang University, Nanchang, ChinaSchool of Information Engineering, Nanchang University, Nanchang, ChinaSchool of Information Engineering, Nanchang University, Nanchang, ChinaSchool of Information Engineering, Nanchang University, Nanchang, ChinaSchool of Information Engineering, Nanchang University, Nanchang, ChinaSchool of Information Engineering, Nanchang University, Nanchang, ChinaSchool of Information Engineering, Nanchang University, Nanchang, ChinaSingle-pixel imaging (SPI), an imaging technique based on the theory of compressed sensing, is limited in real-time imaging and high-resolution images due to its relatively slow imaging speed. In recent years, deep unfolding network compressed sensing reconstruction algorithms based on deep learning have proven to be an effective solution for faster and higher quality image reconstruction. However, existing deep unfolding networks mainly rely on a single piece of a priori information and may ignore other intrinsic structures of the image. Therefore, in this paper, we propose a deep unfolding network (MPDU-Net) that incorporates multiple prior information. To effectively fuse multiple prior information, we propose three different fusion strategies in the deep reconstruction sub-network. An unbiased convolutional layer is used to simulate the sampling reconstruction process to achieve joint reconstruction for effective removal of block artifacts. The sampling matrix is input into the deep reconstruction sub-network as a learnable parameter to achieve joint optimization of sampling reconstruction. Simulation and practical experimental results show that the proposed network outperforms existing compressed sensing reconstruction algorithms based on deep unfolding networks.https://ieeexplore.ieee.org/document/10577085/Single pixel imagingdeep unfolding networkmultiple prior informationjoint optimization
spellingShingle Quan Zou
Qiurong Yan
Qianling Dai
Ao Wang
Bo Yang
Yi Li
Jinwei Yan
Single Pixel Imaging Based on Multiple Prior Deep Unfolding Network
IEEE Photonics Journal
Single pixel imaging
deep unfolding network
multiple prior information
joint optimization
title Single Pixel Imaging Based on Multiple Prior Deep Unfolding Network
title_full Single Pixel Imaging Based on Multiple Prior Deep Unfolding Network
title_fullStr Single Pixel Imaging Based on Multiple Prior Deep Unfolding Network
title_full_unstemmed Single Pixel Imaging Based on Multiple Prior Deep Unfolding Network
title_short Single Pixel Imaging Based on Multiple Prior Deep Unfolding Network
title_sort single pixel imaging based on multiple prior deep unfolding network
topic Single pixel imaging
deep unfolding network
multiple prior information
joint optimization
url https://ieeexplore.ieee.org/document/10577085/
work_keys_str_mv AT quanzou singlepixelimagingbasedonmultiplepriordeepunfoldingnetwork
AT qiurongyan singlepixelimagingbasedonmultiplepriordeepunfoldingnetwork
AT qianlingdai singlepixelimagingbasedonmultiplepriordeepunfoldingnetwork
AT aowang singlepixelimagingbasedonmultiplepriordeepunfoldingnetwork
AT boyang singlepixelimagingbasedonmultiplepriordeepunfoldingnetwork
AT yili singlepixelimagingbasedonmultiplepriordeepunfoldingnetwork
AT jinweiyan singlepixelimagingbasedonmultiplepriordeepunfoldingnetwork