A novel transmission-augmented deep unfolding network with consideration of residual recovery
Abstract Compressive sensing (CS) has been widely applied in signal processing field, especially for image reconstruction tasks. CS simplifies the sampling and compression procedures, but leaves the difficulty to the nonlinear reconstruction. Traditional CS reconstruction algorithms are usually iter...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-01-01
|
| Series: | Complex & Intelligent Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s40747-024-01727-2 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832571191124033536 |
|---|---|
| author | Zhijie Zhang Huang Bai Ljubiša Stanković Junmei Sun Xiumei Li |
| author_facet | Zhijie Zhang Huang Bai Ljubiša Stanković Junmei Sun Xiumei Li |
| author_sort | Zhijie Zhang |
| collection | DOAJ |
| description | Abstract Compressive sensing (CS) has been widely applied in signal processing field, especially for image reconstruction tasks. CS simplifies the sampling and compression procedures, but leaves the difficulty to the nonlinear reconstruction. Traditional CS reconstruction algorithms are usually iterative, having a complete theoretical foundation. Nevertheless, these iterative algorithms suffer from the high computational complexity. The fashionable deep network-based methods can achieve high-precision CS reconstruction with satisfactory speed but are short of theoretical analysis and interpretability. To combine the merits of the above two kinds of CS methods, the deep unfolding networks (DUNs) have been developed. In this paper, a novel DUN named supervised transmission-augmented network (SuperTA-Net) is proposed. Based on the framework of our previous work PIPO-Net, the multi-channel transmission strategy is put forward to reduce the influence of critical information loss between modules and improve the reliability of data. Besides, in order to avoid the issues such as high information redundancy and high computational burden when too many channels are set, the attention based supervision scheme is presented to dynamically adjust the weight of each channel and remove the redundant information. Furthermore, noting the difference between the original image and the output of SuperTA-Net, the reinforcement network is developed, where the main component called residual recovery network (RR-Net) is lightweight and can be added to reinforce all kinds of CS reconstruction networks. Experiments on reconstructing CS images demonstrate the effectiveness of the proposed networks. |
| format | Article |
| id | doaj-art-b1c0de8d158c4d1faa44e53a77f43f15 |
| institution | Kabale University |
| issn | 2199-4536 2198-6053 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Springer |
| record_format | Article |
| series | Complex & Intelligent Systems |
| spelling | doaj-art-b1c0de8d158c4d1faa44e53a77f43f152025-02-02T12:50:17ZengSpringerComplex & Intelligent Systems2199-45362198-60532025-01-0111111510.1007/s40747-024-01727-2A novel transmission-augmented deep unfolding network with consideration of residual recoveryZhijie Zhang0Huang Bai1Ljubiša Stanković2Junmei Sun3Xiumei Li4School of Information Science and Technology, Hangzhou Normal UniversitySchool of Information Science and Technology, Hangzhou Normal UniversityFaculty of Electrical Engineering, University of MontenegroSchool of Information Science and Technology, Hangzhou Normal UniversitySchool of Information Science and Technology, Hangzhou Normal UniversityAbstract Compressive sensing (CS) has been widely applied in signal processing field, especially for image reconstruction tasks. CS simplifies the sampling and compression procedures, but leaves the difficulty to the nonlinear reconstruction. Traditional CS reconstruction algorithms are usually iterative, having a complete theoretical foundation. Nevertheless, these iterative algorithms suffer from the high computational complexity. The fashionable deep network-based methods can achieve high-precision CS reconstruction with satisfactory speed but are short of theoretical analysis and interpretability. To combine the merits of the above two kinds of CS methods, the deep unfolding networks (DUNs) have been developed. In this paper, a novel DUN named supervised transmission-augmented network (SuperTA-Net) is proposed. Based on the framework of our previous work PIPO-Net, the multi-channel transmission strategy is put forward to reduce the influence of critical information loss between modules and improve the reliability of data. Besides, in order to avoid the issues such as high information redundancy and high computational burden when too many channels are set, the attention based supervision scheme is presented to dynamically adjust the weight of each channel and remove the redundant information. Furthermore, noting the difference between the original image and the output of SuperTA-Net, the reinforcement network is developed, where the main component called residual recovery network (RR-Net) is lightweight and can be added to reinforce all kinds of CS reconstruction networks. Experiments on reconstructing CS images demonstrate the effectiveness of the proposed networks.https://doi.org/10.1007/s40747-024-01727-2Compressive sensingDeep unfolding networkMulti-channel transmissionAttention based supervisionResidual recoveryAlternating optimization |
| spellingShingle | Zhijie Zhang Huang Bai Ljubiša Stanković Junmei Sun Xiumei Li A novel transmission-augmented deep unfolding network with consideration of residual recovery Complex & Intelligent Systems Compressive sensing Deep unfolding network Multi-channel transmission Attention based supervision Residual recovery Alternating optimization |
| title | A novel transmission-augmented deep unfolding network with consideration of residual recovery |
| title_full | A novel transmission-augmented deep unfolding network with consideration of residual recovery |
| title_fullStr | A novel transmission-augmented deep unfolding network with consideration of residual recovery |
| title_full_unstemmed | A novel transmission-augmented deep unfolding network with consideration of residual recovery |
| title_short | A novel transmission-augmented deep unfolding network with consideration of residual recovery |
| title_sort | novel transmission augmented deep unfolding network with consideration of residual recovery |
| topic | Compressive sensing Deep unfolding network Multi-channel transmission Attention based supervision Residual recovery Alternating optimization |
| url | https://doi.org/10.1007/s40747-024-01727-2 |
| work_keys_str_mv | AT zhijiezhang anoveltransmissionaugmenteddeepunfoldingnetworkwithconsiderationofresidualrecovery AT huangbai anoveltransmissionaugmenteddeepunfoldingnetworkwithconsiderationofresidualrecovery AT ljubisastankovic anoveltransmissionaugmenteddeepunfoldingnetworkwithconsiderationofresidualrecovery AT junmeisun anoveltransmissionaugmenteddeepunfoldingnetworkwithconsiderationofresidualrecovery AT xiumeili anoveltransmissionaugmenteddeepunfoldingnetworkwithconsiderationofresidualrecovery AT zhijiezhang noveltransmissionaugmenteddeepunfoldingnetworkwithconsiderationofresidualrecovery AT huangbai noveltransmissionaugmenteddeepunfoldingnetworkwithconsiderationofresidualrecovery AT ljubisastankovic noveltransmissionaugmenteddeepunfoldingnetworkwithconsiderationofresidualrecovery AT junmeisun noveltransmissionaugmenteddeepunfoldingnetworkwithconsiderationofresidualrecovery AT xiumeili noveltransmissionaugmenteddeepunfoldingnetworkwithconsiderationofresidualrecovery |