Automatic target recognition of SAR images based on Transformer

Sen Li; Ping Lang; Xiongjun Fu; Jiahuan Jiang; Jian Dong; Zhengang Nie

doi:10.1109/Radar53847.2021.10028126

Automatic target recognition of SAR images based on Transformer

Sen Li, Ping Lang, Xiongjun Fu^*, Jiahuan Jiang, Jian Dong, Zhengang Nie

^*Corresponding author for this work

School of Integrated Circuits and Electronics

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

2 Citations (Scopus)

Abstract

In order to solve the overfitting problem of synthetic aperture radar automatic target recognition (SAR-ATR) at small samples scenarios, this paper proposes a non-subsampling laplace pyramid decomposition (NSLP) based vision transformer (VIT) model, namely NSLP-VIT, for SAR-ATR task. First, the SAR images are preprocessed by feature decomposition via NSLP; Second, a three-channel image is merged as the input of the network using three feature sub-maps; Finally, the preprocessed SAR images are as inputs for VIT network training. The experimental results on Moving and Stationary Target Acquisition and Recognition (MSTAR) data sets show that the proposed method can effectively solve the overfitting problem of SAR-ATR when the data is scarce.

Original language	English
Title of host publication	2021 CIE International Conference on Radar, Radar 2021
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	938-941
Number of pages	4
ISBN (Electronic)	9781665498142
DOIs	https://doi.org/10.1109/Radar53847.2021.10028126
Publication status	Published - 2021
Event	2021 CIE International Conference on Radar, Radar 2021 - Haikou, Hainan, China Duration: 15 Dec 2021 → 19 Dec 2021

Publication series

Name	Proceedings of the IEEE Radar Conference
Volume	2021-December
ISSN (Print)	1097-5764
ISSN (Electronic)	2375-5318

Conference

Conference	2021 CIE International Conference on Radar, Radar 2021
Country/Territory	China
City	Haikou, Hainan
Period	15/12/21 → 19/12/21

Keywords

automatic target recognition
non-subsampling laplace pyramid decomposition
synthetic aperture radar
vision transformer

Access to Document

10.1109/Radar53847.2021.10028126

Cite this

Li, S., Lang, P., Fu, X., Jiang, J., Dong, J., & Nie, Z. (2021). Automatic target recognition of SAR images based on Transformer. In 2021 CIE International Conference on Radar, Radar 2021 (pp. 938-941). (Proceedings of the IEEE Radar Conference; Vol. 2021-December). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/Radar53847.2021.10028126

@inproceedings{689cee835cd24a66b58401ca3834ed50,

title = "Automatic target recognition of SAR images based on Transformer",

abstract = "In order to solve the overfitting problem of synthetic aperture radar automatic target recognition (SAR-ATR) at small samples scenarios, this paper proposes a non-subsampling laplace pyramid decomposition (NSLP) based vision transformer (VIT) model, namely NSLP-VIT, for SAR-ATR task. First, the SAR images are preprocessed by feature decomposition via NSLP; Second, a three-channel image is merged as the input of the network using three feature sub-maps; Finally, the preprocessed SAR images are as inputs for VIT network training. The experimental results on Moving and Stationary Target Acquisition and Recognition (MSTAR) data sets show that the proposed method can effectively solve the overfitting problem of SAR-ATR when the data is scarce.",

keywords = "automatic target recognition, non-subsampling laplace pyramid decomposition, synthetic aperture radar, vision transformer",

author = "Sen Li and Ping Lang and Xiongjun Fu and Jiahuan Jiang and Jian Dong and Zhengang Nie",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 2021 CIE International Conference on Radar, Radar 2021 ; Conference date: 15-12-2021 Through 19-12-2021",

year = "2021",

doi = "10.1109/Radar53847.2021.10028126",

language = "English",

series = "Proceedings of the IEEE Radar Conference",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "938--941",

booktitle = "2021 CIE International Conference on Radar, Radar 2021",

address = "United States",

}

Li, S, Lang, P, Fu, X, Jiang, J, Dong, J & Nie, Z 2021, Automatic target recognition of SAR images based on Transformer. in 2021 CIE International Conference on Radar, Radar 2021. Proceedings of the IEEE Radar Conference, vol. 2021-December, Institute of Electrical and Electronics Engineers Inc., pp. 938-941, 2021 CIE International Conference on Radar, Radar 2021, Haikou, Hainan, China, 15/12/21. https://doi.org/10.1109/Radar53847.2021.10028126

Automatic target recognition of SAR images based on Transformer. / Li, Sen; Lang, Ping; Fu, Xiongjun et al.
2021 CIE International Conference on Radar, Radar 2021. Institute of Electrical and Electronics Engineers Inc., 2021. p. 938-941 (Proceedings of the IEEE Radar Conference; Vol. 2021-December).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Automatic target recognition of SAR images based on Transformer

AU - Li, Sen

AU - Lang, Ping

AU - Fu, Xiongjun

AU - Jiang, Jiahuan

AU - Dong, Jian

AU - Nie, Zhengang

PY - 2021

Y1 - 2021

N2 - In order to solve the overfitting problem of synthetic aperture radar automatic target recognition (SAR-ATR) at small samples scenarios, this paper proposes a non-subsampling laplace pyramid decomposition (NSLP) based vision transformer (VIT) model, namely NSLP-VIT, for SAR-ATR task. First, the SAR images are preprocessed by feature decomposition via NSLP; Second, a three-channel image is merged as the input of the network using three feature sub-maps; Finally, the preprocessed SAR images are as inputs for VIT network training. The experimental results on Moving and Stationary Target Acquisition and Recognition (MSTAR) data sets show that the proposed method can effectively solve the overfitting problem of SAR-ATR when the data is scarce.

AB - In order to solve the overfitting problem of synthetic aperture radar automatic target recognition (SAR-ATR) at small samples scenarios, this paper proposes a non-subsampling laplace pyramid decomposition (NSLP) based vision transformer (VIT) model, namely NSLP-VIT, for SAR-ATR task. First, the SAR images are preprocessed by feature decomposition via NSLP; Second, a three-channel image is merged as the input of the network using three feature sub-maps; Finally, the preprocessed SAR images are as inputs for VIT network training. The experimental results on Moving and Stationary Target Acquisition and Recognition (MSTAR) data sets show that the proposed method can effectively solve the overfitting problem of SAR-ATR when the data is scarce.

KW - automatic target recognition

KW - non-subsampling laplace pyramid decomposition

KW - synthetic aperture radar

KW - vision transformer

UR - http://www.scopus.com/inward/record.url?scp=85181065351&partnerID=8YFLogxK

U2 - 10.1109/Radar53847.2021.10028126

DO - 10.1109/Radar53847.2021.10028126

M3 - Conference contribution

AN - SCOPUS:85181065351

T3 - Proceedings of the IEEE Radar Conference

SP - 938

EP - 941

BT - 2021 CIE International Conference on Radar, Radar 2021

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2021 CIE International Conference on Radar, Radar 2021

Y2 - 15 December 2021 through 19 December 2021

ER -

Automatic target recognition of SAR images based on Transformer

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this