Diffusion Model With Gradient Descent Module Guiding Reconstruction for Single-Pixel Imaging
Reconstructing high-quality images with few measurements has always been a primary goal for single-pixel imaging (SPI). Diffusion models have shown outstanding performance in image generation and have been effectively attempted in image reconstruction for ghost imaging. However, there is still a gre...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Photonics Journal |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10613365/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850093740966805504 |
|---|---|
| author | Chen Huang Qiurong Yan Jinwei Yan Yi Li Xiaolong Luo Hui Wang |
| author_facet | Chen Huang Qiurong Yan Jinwei Yan Yi Li Xiaolong Luo Hui Wang |
| author_sort | Chen Huang |
| collection | DOAJ |
| description | Reconstructing high-quality images with few measurements has always been a primary goal for single-pixel imaging (SPI). Diffusion models have shown outstanding performance in image generation and have been effectively attempted in image reconstruction for ghost imaging. However, there is still a great deal of space for improvement in the quality of image reconstruction at low sampling rates. Inspired by the proximal gradient descent algorithm (PGD), we propose Diffusion Model with Gradient Descent Module Guiding Reconstruction for Single-Pixel Imaging. The gradient descent module in PGD is utilized for preliminary image reconstruction. The preliminary reconstruction serves as prior information to iteratively constrain the diffusion model, allowing it to generate target images consistent with the training data distribution. Additionally, the strong mapping ability of the diffusion model replaces the traditional proximal operator to accelerate convergence. Full connected sampling and convolutional sampling are proposed as alternative sampling methods to the traditional Gaussian random matrix sampling. Sampling and generation are optimized jointly to capture key image information and improve reconstruction accuracy. Simulations and experiments confirm that our proposed network can significantly improve the quality of image reconstruction at low measurement rates. |
| format | Article |
| id | doaj-art-add6b4d0a3004e6ea80b6afe2a4ad4e5 |
| institution | DOAJ |
| issn | 1943-0655 |
| language | English |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Photonics Journal |
| spelling | doaj-art-add6b4d0a3004e6ea80b6afe2a4ad4e52025-08-20T02:41:49ZengIEEEIEEE Photonics Journal1943-06552024-01-0116411010.1109/JPHOT.2024.343497210613365Diffusion Model With Gradient Descent Module Guiding Reconstruction for Single-Pixel ImagingChen Huang0Qiurong Yan1https://orcid.org/0000-0003-4736-7435Jinwei Yan2Yi Li3Xiaolong Luo4Hui Wang5School 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, ChinaCollege of Physics and Communication Electronics, Jiangxi Normal University, Nanchang, ChinaReconstructing high-quality images with few measurements has always been a primary goal for single-pixel imaging (SPI). Diffusion models have shown outstanding performance in image generation and have been effectively attempted in image reconstruction for ghost imaging. However, there is still a great deal of space for improvement in the quality of image reconstruction at low sampling rates. Inspired by the proximal gradient descent algorithm (PGD), we propose Diffusion Model with Gradient Descent Module Guiding Reconstruction for Single-Pixel Imaging. The gradient descent module in PGD is utilized for preliminary image reconstruction. The preliminary reconstruction serves as prior information to iteratively constrain the diffusion model, allowing it to generate target images consistent with the training data distribution. Additionally, the strong mapping ability of the diffusion model replaces the traditional proximal operator to accelerate convergence. Full connected sampling and convolutional sampling are proposed as alternative sampling methods to the traditional Gaussian random matrix sampling. Sampling and generation are optimized jointly to capture key image information and improve reconstruction accuracy. Simulations and experiments confirm that our proposed network can significantly improve the quality of image reconstruction at low measurement rates.https://ieeexplore.ieee.org/document/10613365/Compressed sensing (CS)single pixel imaging (SPI)diffusion models(DMs)proximal gradient descent |
| spellingShingle | Chen Huang Qiurong Yan Jinwei Yan Yi Li Xiaolong Luo Hui Wang Diffusion Model With Gradient Descent Module Guiding Reconstruction for Single-Pixel Imaging IEEE Photonics Journal Compressed sensing (CS) single pixel imaging (SPI) diffusion models(DMs) proximal gradient descent |
| title | Diffusion Model With Gradient Descent Module Guiding Reconstruction for Single-Pixel Imaging |
| title_full | Diffusion Model With Gradient Descent Module Guiding Reconstruction for Single-Pixel Imaging |
| title_fullStr | Diffusion Model With Gradient Descent Module Guiding Reconstruction for Single-Pixel Imaging |
| title_full_unstemmed | Diffusion Model With Gradient Descent Module Guiding Reconstruction for Single-Pixel Imaging |
| title_short | Diffusion Model With Gradient Descent Module Guiding Reconstruction for Single-Pixel Imaging |
| title_sort | diffusion model with gradient descent module guiding reconstruction for single pixel imaging |
| topic | Compressed sensing (CS) single pixel imaging (SPI) diffusion models(DMs) proximal gradient descent |
| url | https://ieeexplore.ieee.org/document/10613365/ |
| work_keys_str_mv | AT chenhuang diffusionmodelwithgradientdescentmoduleguidingreconstructionforsinglepixelimaging AT qiurongyan diffusionmodelwithgradientdescentmoduleguidingreconstructionforsinglepixelimaging AT jinweiyan diffusionmodelwithgradientdescentmoduleguidingreconstructionforsinglepixelimaging AT yili diffusionmodelwithgradientdescentmoduleguidingreconstructionforsinglepixelimaging AT xiaolongluo diffusionmodelwithgradientdescentmoduleguidingreconstructionforsinglepixelimaging AT huiwang diffusionmodelwithgradientdescentmoduleguidingreconstructionforsinglepixelimaging |