SAR Image Despeckling Based on Block-Matching and Noise-Referenced Deep Learning Method
The noise2noise-based despeckling method, capable of training the despeckling deep neural network with only noisy synthetic aperture radar (SAR) image, has presented very good performance in recent research. This method requires a fine-registered multi-temporal dataset with minor time variance and u...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-02-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/14/4/931 |
| _version_ | 1827652950446571520 |
|---|---|
| author | Chen Wang Zhixiang Yin Xiaoshuang Ma Zhutao Yang |
| author_facet | Chen Wang Zhixiang Yin Xiaoshuang Ma Zhutao Yang |
| author_sort | Chen Wang |
| collection | DOAJ |
| description | The noise2noise-based despeckling method, capable of training the despeckling deep neural network with only noisy synthetic aperture radar (SAR) image, has presented very good performance in recent research. This method requires a fine-registered multi-temporal dataset with minor time variance and uses similarity estimation to compensate for the time variance. However, constructing such a training dataset is very time-consuming and may not be viable for a certain practitioner. In this article, we propose a novel single-image-capable speckling method that combines the similarity-based block-matching and noise referenced deep learning network. The denoising network designed for this method is an encoder–decoder convolutional neural network and is accommodated to small image patches. This method firstly constructs a large number of noisy pairs as training input by similarity-based block-matching in either one noisy SAR image or multiple images. Then, the method trains the network in a Siamese manner with two parameter-sharing branches. The proposed method demonstrates favorable despeckling performance with both simulated and real SAR data with respect to other state-of-the-art reference filters. It also presents satisfying generalization capability as the trained network can despeckle well the unseen image of the same sensor. The main advantage of the proposed method is its application flexibility. It could be trained with either one noisy image or multiple images. Furthermore, the despeckling could be inferred by either the ad hoc trained network or a pre-trained one of the same sensor. |
| first_indexed | 2024-03-09T21:09:18Z |
| format | Article |
| id | doaj.art-88a1505502e344edb9350d1c433779be |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-09T21:09:18Z |
| publishDate | 2022-02-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-88a1505502e344edb9350d1c433779be2023-11-23T21:54:20ZengMDPI AGRemote Sensing2072-42922022-02-0114493110.3390/rs14040931SAR Image Despeckling Based on Block-Matching and Noise-Referenced Deep Learning MethodChen Wang0Zhixiang Yin1Xiaoshuang Ma2Zhutao Yang3School of Resources and Environmental Engineering, Anhui University, Hefei 230601, ChinaSchool of Resources and Environmental Engineering, Anhui University, Hefei 230601, ChinaSchool of Resources and Environmental Engineering, Anhui University, Hefei 230601, ChinaCentral Southern China Electric Power Design Institute Co., Ltd., Wuhan 430071, ChinaThe noise2noise-based despeckling method, capable of training the despeckling deep neural network with only noisy synthetic aperture radar (SAR) image, has presented very good performance in recent research. This method requires a fine-registered multi-temporal dataset with minor time variance and uses similarity estimation to compensate for the time variance. However, constructing such a training dataset is very time-consuming and may not be viable for a certain practitioner. In this article, we propose a novel single-image-capable speckling method that combines the similarity-based block-matching and noise referenced deep learning network. The denoising network designed for this method is an encoder–decoder convolutional neural network and is accommodated to small image patches. This method firstly constructs a large number of noisy pairs as training input by similarity-based block-matching in either one noisy SAR image or multiple images. Then, the method trains the network in a Siamese manner with two parameter-sharing branches. The proposed method demonstrates favorable despeckling performance with both simulated and real SAR data with respect to other state-of-the-art reference filters. It also presents satisfying generalization capability as the trained network can despeckle well the unseen image of the same sensor. The main advantage of the proposed method is its application flexibility. It could be trained with either one noisy image or multiple images. Furthermore, the despeckling could be inferred by either the ad hoc trained network or a pre-trained one of the same sensor.https://www.mdpi.com/2072-4292/14/4/931block matchingconvolutional neural network (CNN)deep learningspeckle filteringsynthetic aperture radar (SAR)image similarity |
| spellingShingle | Chen Wang Zhixiang Yin Xiaoshuang Ma Zhutao Yang SAR Image Despeckling Based on Block-Matching and Noise-Referenced Deep Learning Method Remote Sensing block matching convolutional neural network (CNN) deep learning speckle filtering synthetic aperture radar (SAR) image similarity |
| title | SAR Image Despeckling Based on Block-Matching and Noise-Referenced Deep Learning Method |
| title_full | SAR Image Despeckling Based on Block-Matching and Noise-Referenced Deep Learning Method |
| title_fullStr | SAR Image Despeckling Based on Block-Matching and Noise-Referenced Deep Learning Method |
| title_full_unstemmed | SAR Image Despeckling Based on Block-Matching and Noise-Referenced Deep Learning Method |
| title_short | SAR Image Despeckling Based on Block-Matching and Noise-Referenced Deep Learning Method |
| title_sort | sar image despeckling based on block matching and noise referenced deep learning method |
| topic | block matching convolutional neural network (CNN) deep learning speckle filtering synthetic aperture radar (SAR) image similarity |
| url | https://www.mdpi.com/2072-4292/14/4/931 |
| work_keys_str_mv | AT chenwang sarimagedespecklingbasedonblockmatchingandnoisereferenceddeeplearningmethod AT zhixiangyin sarimagedespecklingbasedonblockmatchingandnoisereferenceddeeplearningmethod AT xiaoshuangma sarimagedespecklingbasedonblockmatchingandnoisereferenceddeeplearningmethod AT zhutaoyang sarimagedespecklingbasedonblockmatchingandnoisereferenceddeeplearningmethod |