Pixel-wise segmentation of SAR imagery using encoder-decoder network and fully-connected CRF

Fei Gao; Yishan He; Jun Wang; Fei Ma; Erfu Yang; Amir Hussain

doi:10.1007/978-3-030-39431-8_15

Pixel-wise segmentation of SAR imagery using encoder-decoder network and fully-connected CRF

Fei Gao, Yishan He^*, Jun Wang, Fei Ma, Erfu Yang, Amir Hussain

^*Corresponding author for this work

Design, Manufacturing And Engineering Management

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution book

37 Downloads (Pure)

Abstract

Synthetic Aperture Radar (SAR) image segmentation is an important step in SAR image interpretation. Common Patch-based methods treat all the pixels within the patch as a single category and do not take the label consistency between neighbor patches into consideration, which makes the segmentation results less accurate. In this paper, we use an encoder-decoder network to conduct pixel-wise segmentation. Then, in order to make full use of the contextual information between patches, we use fully-connected conditional random field to optimize the combined probability map output from encoder-decoder network. The testing results on our SAR data set shows that our method can effectively maintain contextual information of pixels and achieve better segmentation results.

Original language	English
Title of host publication	Advances in Brain Inspired Cognitive Systems
Subtitle of host publication	10th International Conference, BICS 2019, Guangzhou, China, July 13–14, 2019, Proceedings
Editors	Jinchang Ren, Amir Hussain, Huimin Zhao, Kaizhu Huang, Jiangbin Zheng, Jun Cai, Rongjun Chen, Yinyin Xiao
Place of Publication	Cham, Switzerland
Publisher	Springer
Pages	155-165
Number of pages	11
ISBN (Print)	9783030394301
DOIs	https://doi.org/10.1007/978-3-030-39431-8_15
Publication status	Published - 1 Feb 2020
Event	10th International Conference on Brain Inspired Cognitive Systems, BICS 2019 - Guangzhou, China Duration: 13 Jul 2019 → 14 Jul 2019

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11691 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	10th International Conference on Brain Inspired Cognitive Systems, BICS 2019
Country/Territory	China
City	Guangzhou
Period	13/07/19 → 14/07/19

Keywords

encoder-decoder network
fully-connected CRF
SAR image segmentation

Access to Document

10.1007/978-3-030-39431-8_15

Gao-etal-BICS2019-Pixel-wise-segmentation-of-SAR-imagery-using-encoder-decoder-networkAccepted author manuscript, 610 KB

Cite this

Gao, F., He, Y., Wang, J., Ma, F., Yang, E., & Hussain, A. (2020). Pixel-wise segmentation of SAR imagery using encoder-decoder network and fully-connected CRF. In J. Ren, A. Hussain, H. Zhao, K. Huang, J. Zheng, J. Cai, R. Chen, & Y. Xiao (Eds.), Advances in Brain Inspired Cognitive Systems: 10th International Conference, BICS 2019, Guangzhou, China, July 13–14, 2019, Proceedings (pp. 155-165). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11691 LNAI). Springer. https://doi.org/10.1007/978-3-030-39431-8_15

Gao, Fei ; He, Yishan ; Wang, Jun et al. / Pixel-wise segmentation of SAR imagery using encoder-decoder network and fully-connected CRF. Advances in Brain Inspired Cognitive Systems: 10th International Conference, BICS 2019, Guangzhou, China, July 13–14, 2019, Proceedings. editor / Jinchang Ren ; Amir Hussain ; Huimin Zhao ; Kaizhu Huang ; Jiangbin Zheng ; Jun Cai ; Rongjun Chen ; Yinyin Xiao. Cham, Switzerland : Springer, 2020. pp. 155-165 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{511948af8f214640a5ded0f5b730d20e,

title = "Pixel-wise segmentation of SAR imagery using encoder-decoder network and fully-connected CRF",

abstract = "Synthetic Aperture Radar (SAR) image segmentation is an important step in SAR image interpretation. Common Patch-based methods treat all the pixels within the patch as a single category and do not take the label consistency between neighbor patches into consideration, which makes the segmentation results less accurate. In this paper, we use an encoder-decoder network to conduct pixel-wise segmentation. Then, in order to make full use of the contextual information between patches, we use fully-connected conditional random field to optimize the combined probability map output from encoder-decoder network. The testing results on our SAR data set shows that our method can effectively maintain contextual information of pixels and achieve better segmentation results.",

keywords = "encoder-decoder network, fully-connected CRF, SAR image segmentation",

author = "Fei Gao and Yishan He and Jun Wang and Fei Ma and Erfu Yang and Amir Hussain",

year = "2020",

month = feb,

day = "1",

doi = "10.1007/978-3-030-39431-8_15",

language = "English",

isbn = "9783030394301",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "155--165",

editor = "Jinchang Ren and Amir Hussain and Huimin Zhao and Kaizhu Huang and Jiangbin Zheng and Jun Cai and Rongjun Chen and Yinyin Xiao",

booktitle = "Advances in Brain Inspired Cognitive Systems",

note = "10th International Conference on Brain Inspired Cognitive Systems, BICS 2019 ; Conference date: 13-07-2019 Through 14-07-2019",

}

Gao, F, He, Y, Wang, J, Ma, F, Yang, E & Hussain, A 2020, Pixel-wise segmentation of SAR imagery using encoder-decoder network and fully-connected CRF. in J Ren, A Hussain, H Zhao, K Huang, J Zheng, J Cai, R Chen & Y Xiao (eds), Advances in Brain Inspired Cognitive Systems: 10th International Conference, BICS 2019, Guangzhou, China, July 13–14, 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11691 LNAI, Springer, Cham, Switzerland, pp. 155-165, 10th International Conference on Brain Inspired Cognitive Systems, BICS 2019, Guangzhou, China, 13/07/19. https://doi.org/10.1007/978-3-030-39431-8_15

Pixel-wise segmentation of SAR imagery using encoder-decoder network and fully-connected CRF. / Gao, Fei; He, Yishan; Wang, Jun et al.
Advances in Brain Inspired Cognitive Systems: 10th International Conference, BICS 2019, Guangzhou, China, July 13–14, 2019, Proceedings. ed. / Jinchang Ren; Amir Hussain; Huimin Zhao; Kaizhu Huang; Jiangbin Zheng; Jun Cai; Rongjun Chen; Yinyin Xiao. Cham, Switzerland: Springer, 2020. p. 155-165 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11691 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution book

TY - GEN

T1 - Pixel-wise segmentation of SAR imagery using encoder-decoder network and fully-connected CRF

AU - Gao, Fei

AU - He, Yishan

AU - Wang, Jun

AU - Ma, Fei

AU - Yang, Erfu

AU - Hussain, Amir

PY - 2020/2/1

Y1 - 2020/2/1

N2 - Synthetic Aperture Radar (SAR) image segmentation is an important step in SAR image interpretation. Common Patch-based methods treat all the pixels within the patch as a single category and do not take the label consistency between neighbor patches into consideration, which makes the segmentation results less accurate. In this paper, we use an encoder-decoder network to conduct pixel-wise segmentation. Then, in order to make full use of the contextual information between patches, we use fully-connected conditional random field to optimize the combined probability map output from encoder-decoder network. The testing results on our SAR data set shows that our method can effectively maintain contextual information of pixels and achieve better segmentation results.

AB - Synthetic Aperture Radar (SAR) image segmentation is an important step in SAR image interpretation. Common Patch-based methods treat all the pixels within the patch as a single category and do not take the label consistency between neighbor patches into consideration, which makes the segmentation results less accurate. In this paper, we use an encoder-decoder network to conduct pixel-wise segmentation. Then, in order to make full use of the contextual information between patches, we use fully-connected conditional random field to optimize the combined probability map output from encoder-decoder network. The testing results on our SAR data set shows that our method can effectively maintain contextual information of pixels and achieve better segmentation results.

KW - encoder-decoder network

KW - fully-connected CRF

KW - SAR image segmentation

UR - http://www.scopus.com/inward/record.url?scp=85080961868&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-39431-8_15

DO - 10.1007/978-3-030-39431-8_15

M3 - Conference contribution book

AN - SCOPUS:85080961868

SN - 9783030394301

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 155

EP - 165

BT - Advances in Brain Inspired Cognitive Systems

A2 - Ren, Jinchang

A2 - Hussain, Amir

A2 - Zhao, Huimin

A2 - Huang, Kaizhu

A2 - Zheng, Jiangbin

A2 - Cai, Jun

A2 - Chen, Rongjun

A2 - Xiao, Yinyin

PB - Springer

CY - Cham, Switzerland

T2 - 10th International Conference on Brain Inspired Cognitive Systems, BICS 2019

Y2 - 13 July 2019 through 14 July 2019

ER -

Gao F, He Y, Wang J, Ma F, Yang E, Hussain A. Pixel-wise segmentation of SAR imagery using encoder-decoder network and fully-connected CRF. In Ren J, Hussain A, Zhao H, Huang K, Zheng J, Cai J, Chen R, Xiao Y, editors, Advances in Brain Inspired Cognitive Systems: 10th International Conference, BICS 2019, Guangzhou, China, July 13–14, 2019, Proceedings. Cham, Switzerland: Springer. 2020. p. 155-165. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-39431-8_15

Pixel-wise segmentation of SAR imagery using encoder-decoder network and fully-connected CRF

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this