Improved Sea Ice Image Segmentation Using U 2 -Net and Dataset Augmentation
Sea ice extraction and segmentation of remote sensing images is the basis for sea ice monitoring. Traditional image segmentation methods rely on manual sampling and require complex feature extraction. Deep-learning-based semantic segmentation methods have the advantages of high efficiency, intellige...
Published in: | Applied Sciences |
---|---|
Main Authors: | , , , , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
MDPI AG
2023
|
Subjects: | |
Online Access: | https://doi.org/10.3390/app13169402 https://doaj.org/article/50fdd993dd7a43b7b2125ce697d0e8b5 |
id |
ftdoajarticles:oai:doaj.org/article:50fdd993dd7a43b7b2125ce697d0e8b5 |
---|---|
record_format |
openpolar |
spelling |
ftdoajarticles:oai:doaj.org/article:50fdd993dd7a43b7b2125ce697d0e8b5 2023-09-26T15:22:54+02:00 Improved Sea Ice Image Segmentation Using U 2 -Net and Dataset Augmentation Yongjian Li He Li Dazhao Fan Zhixin Li Song Ji 2023-08-01T00:00:00Z https://doi.org/10.3390/app13169402 https://doaj.org/article/50fdd993dd7a43b7b2125ce697d0e8b5 EN eng MDPI AG https://www.mdpi.com/2076-3417/13/16/9402 https://doaj.org/toc/2076-3417 doi:10.3390/app13169402 2076-3417 https://doaj.org/article/50fdd993dd7a43b7b2125ce697d0e8b5 Applied Sciences, Vol 13, Iss 9402, p 9402 (2023) sea ice segmentation U 2 -Net remote sensing images Technology T Engineering (General). Civil engineering (General) TA1-2040 Biology (General) QH301-705.5 Physics QC1-999 Chemistry QD1-999 article 2023 ftdoajarticles https://doi.org/10.3390/app13169402 2023-08-27T00:36:09Z Sea ice extraction and segmentation of remote sensing images is the basis for sea ice monitoring. Traditional image segmentation methods rely on manual sampling and require complex feature extraction. Deep-learning-based semantic segmentation methods have the advantages of high efficiency, intelligence, and automation. Sea ice segmentation using deep learning methods faces the following problems: in terms of datasets, the high cost of sea ice image label production leads to fewer datasets for sea ice segmentation; in terms of image quality, remote sensing image noise and severe weather conditions affect image quality, which affects the accuracy of sea ice extraction. To address the quantity and quality of the dataset, this study used multiple data augmentation methods for data expansion. To improve the semantic segmentation accuracy, the SC-U 2 -Net network was constructed using multiscale inflation convolution and a multilayer convolutional block attention module (CBAM) attention mechanism for the U 2 -Net network. The experiments showed that (1) data augmentation solved the problem of an insufficient number of training samples to a certain extent and improved the accuracy of image segmentation; (2) this study designed a multilevel Gaussian noise data augmentation scheme to improve the network’s ability to resist noise interference and achieve a more accurate segmentation of images with different degrees of noise pollution; (3) the inclusion of a multiscale inflation perceptron and multilayer CBAM attention mechanism improved the ability of U 2 -Net network feature extraction and enhanced the model accuracy and generalization ability. Article in Journal/Newspaper Sea ice Directory of Open Access Journals: DOAJ Articles Applied Sciences 13 16 9402 |
institution |
Open Polar |
collection |
Directory of Open Access Journals: DOAJ Articles |
op_collection_id |
ftdoajarticles |
language |
English |
topic |
sea ice segmentation U 2 -Net remote sensing images Technology T Engineering (General). Civil engineering (General) TA1-2040 Biology (General) QH301-705.5 Physics QC1-999 Chemistry QD1-999 |
spellingShingle |
sea ice segmentation U 2 -Net remote sensing images Technology T Engineering (General). Civil engineering (General) TA1-2040 Biology (General) QH301-705.5 Physics QC1-999 Chemistry QD1-999 Yongjian Li He Li Dazhao Fan Zhixin Li Song Ji Improved Sea Ice Image Segmentation Using U 2 -Net and Dataset Augmentation |
topic_facet |
sea ice segmentation U 2 -Net remote sensing images Technology T Engineering (General). Civil engineering (General) TA1-2040 Biology (General) QH301-705.5 Physics QC1-999 Chemistry QD1-999 |
description |
Sea ice extraction and segmentation of remote sensing images is the basis for sea ice monitoring. Traditional image segmentation methods rely on manual sampling and require complex feature extraction. Deep-learning-based semantic segmentation methods have the advantages of high efficiency, intelligence, and automation. Sea ice segmentation using deep learning methods faces the following problems: in terms of datasets, the high cost of sea ice image label production leads to fewer datasets for sea ice segmentation; in terms of image quality, remote sensing image noise and severe weather conditions affect image quality, which affects the accuracy of sea ice extraction. To address the quantity and quality of the dataset, this study used multiple data augmentation methods for data expansion. To improve the semantic segmentation accuracy, the SC-U 2 -Net network was constructed using multiscale inflation convolution and a multilayer convolutional block attention module (CBAM) attention mechanism for the U 2 -Net network. The experiments showed that (1) data augmentation solved the problem of an insufficient number of training samples to a certain extent and improved the accuracy of image segmentation; (2) this study designed a multilevel Gaussian noise data augmentation scheme to improve the network’s ability to resist noise interference and achieve a more accurate segmentation of images with different degrees of noise pollution; (3) the inclusion of a multiscale inflation perceptron and multilayer CBAM attention mechanism improved the ability of U 2 -Net network feature extraction and enhanced the model accuracy and generalization ability. |
format |
Article in Journal/Newspaper |
author |
Yongjian Li He Li Dazhao Fan Zhixin Li Song Ji |
author_facet |
Yongjian Li He Li Dazhao Fan Zhixin Li Song Ji |
author_sort |
Yongjian Li |
title |
Improved Sea Ice Image Segmentation Using U 2 -Net and Dataset Augmentation |
title_short |
Improved Sea Ice Image Segmentation Using U 2 -Net and Dataset Augmentation |
title_full |
Improved Sea Ice Image Segmentation Using U 2 -Net and Dataset Augmentation |
title_fullStr |
Improved Sea Ice Image Segmentation Using U 2 -Net and Dataset Augmentation |
title_full_unstemmed |
Improved Sea Ice Image Segmentation Using U 2 -Net and Dataset Augmentation |
title_sort |
improved sea ice image segmentation using u 2 -net and dataset augmentation |
publisher |
MDPI AG |
publishDate |
2023 |
url |
https://doi.org/10.3390/app13169402 https://doaj.org/article/50fdd993dd7a43b7b2125ce697d0e8b5 |
genre |
Sea ice |
genre_facet |
Sea ice |
op_source |
Applied Sciences, Vol 13, Iss 9402, p 9402 (2023) |
op_relation |
https://www.mdpi.com/2076-3417/13/16/9402 https://doaj.org/toc/2076-3417 doi:10.3390/app13169402 2076-3417 https://doaj.org/article/50fdd993dd7a43b7b2125ce697d0e8b5 |
op_doi |
https://doi.org/10.3390/app13169402 |
container_title |
Applied Sciences |
container_volume |
13 |
container_issue |
16 |
container_start_page |
9402 |
_version_ |
1778148830958583808 |