Unsupervised Hyperspectral Denoising Based on Deep Image Prior and Least Favorable Distribution
This article considers the inverse problem under hyperspectral images (HSIs) denoising framework. Recently, it has been shown that deep learning is a promising approach to image denoising. However, deep learning to be effective usually needs a massive amount of training data. Moreover, in a practica...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9813381/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832582414475460608 |
|---|---|
| author | Keivan Faghih Niresi Chong-Yung Chi |
| author_facet | Keivan Faghih Niresi Chong-Yung Chi |
| author_sort | Keivan Faghih Niresi |
| collection | DOAJ |
| description | This article considers the inverse problem under hyperspectral images (HSIs) denoising framework. Recently, it has been shown that deep learning is a promising approach to image denoising. However, deep learning to be effective usually needs a massive amount of training data. Moreover, in a practical scenario, HSIs may get contaminated by different kinds of noises such as Gaussian and/or sparse noise. Lately, it has been reported that the convolutional neural network (CNN), the core element used by deep image prior (DIP), is able to capture image statistical characteristics without the need of training, i.e., restore the clean image blindly. Nonetheless, there exists some performance gap between DIP and state-of-the-art methods in HSIs (e.g., low-rank models). By applying the Huber loss function (HLF), which is derived through a least favorable distribution in robust statistics, to DIP, we propose a novel unsupervised denoising algorithm, referred as to the HLF-DIP, free from pretraining and without involving any regularizer. Extensive experimental results are provided to demonstrate that the proposed HLF-DIP algorithm significantly outperforms seven state-of-the-art algorithms in both complexity (thanks to no regularization) and robustness against complex noise (e.g., mixed types of noises). |
| format | Article |
| id | doaj-art-eb68a3e1caea46a8a6dc79aa29d3df79 |
| institution | Kabale University |
| issn | 1939-1404 2151-1535 |
| language | English |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj-art-eb68a3e1caea46a8a6dc79aa29d3df792025-01-30T00:00:10ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352022-01-01155967598310.1109/JSTARS.2022.31877229813381Unsupervised Hyperspectral Denoising Based on Deep Image Prior and Least Favorable DistributionKeivan Faghih Niresi0https://orcid.org/0000-0003-0311-5877Chong-Yung Chi1https://orcid.org/0000-0001-5004-7155Institute of Communications Engineering, National Tsing Hua University, Hsinchu, TaiwanInstitute of Communications Engineering and Department of Electrical Engineering, National Tsing Hua University, Hsinchu, TaiwanThis article considers the inverse problem under hyperspectral images (HSIs) denoising framework. Recently, it has been shown that deep learning is a promising approach to image denoising. However, deep learning to be effective usually needs a massive amount of training data. Moreover, in a practical scenario, HSIs may get contaminated by different kinds of noises such as Gaussian and/or sparse noise. Lately, it has been reported that the convolutional neural network (CNN), the core element used by deep image prior (DIP), is able to capture image statistical characteristics without the need of training, i.e., restore the clean image blindly. Nonetheless, there exists some performance gap between DIP and state-of-the-art methods in HSIs (e.g., low-rank models). By applying the Huber loss function (HLF), which is derived through a least favorable distribution in robust statistics, to DIP, we propose a novel unsupervised denoising algorithm, referred as to the HLF-DIP, free from pretraining and without involving any regularizer. Extensive experimental results are provided to demonstrate that the proposed HLF-DIP algorithm significantly outperforms seven state-of-the-art algorithms in both complexity (thanks to no regularization) and robustness against complex noise (e.g., mixed types of noises).https://ieeexplore.ieee.org/document/9813381/Adam optimizerconvolutional neural networkdeep image priorHuber loss functionhyperspectral denoisingleast favorable distribution |
| spellingShingle | Keivan Faghih Niresi Chong-Yung Chi Unsupervised Hyperspectral Denoising Based on Deep Image Prior and Least Favorable Distribution IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Adam optimizer convolutional neural network deep image prior Huber loss function hyperspectral denoising least favorable distribution |
| title | Unsupervised Hyperspectral Denoising Based on Deep Image Prior and Least Favorable Distribution |
| title_full | Unsupervised Hyperspectral Denoising Based on Deep Image Prior and Least Favorable Distribution |
| title_fullStr | Unsupervised Hyperspectral Denoising Based on Deep Image Prior and Least Favorable Distribution |
| title_full_unstemmed | Unsupervised Hyperspectral Denoising Based on Deep Image Prior and Least Favorable Distribution |
| title_short | Unsupervised Hyperspectral Denoising Based on Deep Image Prior and Least Favorable Distribution |
| title_sort | unsupervised hyperspectral denoising based on deep image prior and least favorable distribution |
| topic | Adam optimizer convolutional neural network deep image prior Huber loss function hyperspectral denoising least favorable distribution |
| url | https://ieeexplore.ieee.org/document/9813381/ |
| work_keys_str_mv | AT keivanfaghihniresi unsupervisedhyperspectraldenoisingbasedondeepimagepriorandleastfavorabledistribution AT chongyungchi unsupervisedhyperspectraldenoisingbasedondeepimagepriorandleastfavorabledistribution |