Target Recognition Method Based on Graph Structure Perception of Invariant Features for SAR Images

Synthetic Aperture Radar (SAR) image target recognition technology based on deep learning has matured. However, challenges remain due to scattering phenomenon and noise interference that cause significant intraclass variability in imaging results. Invariant features, which represent the essential at...

Full description

Saved in:
Bibliographic Details
Main Authors: Jingyi CAO, Yang ZHANG, Ya’nan YOU, Yamin WANG, Feng YANG, Weijia REN, Jun LIU
Format: Article
Language:English
Published: China Science Publishing & Media Ltd. (CSPM) 2025-04-01
Series:Leida xuebao
Subjects:
Online Access:https://radars.ac.cn/cn/article/doi/10.12000/JR24125
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850050120796602368
author Jingyi CAO
Yang ZHANG
Ya’nan YOU
Yamin WANG
Feng YANG
Weijia REN
Jun LIU
author_facet Jingyi CAO
Yang ZHANG
Ya’nan YOU
Yamin WANG
Feng YANG
Weijia REN
Jun LIU
author_sort Jingyi CAO
collection DOAJ
description Synthetic Aperture Radar (SAR) image target recognition technology based on deep learning has matured. However, challenges remain due to scattering phenomenon and noise interference that cause significant intraclass variability in imaging results. Invariant features, which represent the essential attributes of a specific target class with consistent expressions, are crucial for high-precision recognition. We define these invariant features from the entity, its surrounding environment, and their combined context as the target’s essential features. Guided by multilevel essential feature modeling theory, we propose a SAR image target recognition method based on graph networks and invariant feature perception. This method employs a dual-branch network to process multiview SAR images simultaneously using a rotation-learnable unit to adaptively align dual-branch features and reinforce invariant features with rotational immunity by minimizing intraclass feature differences. Specifically, to support essential feature extraction in each branch, we design a feature-guided graph feature perception module based on multilevel essential feature modeling. This module uses salient points for target feature analysis and comprises a target ontology feature enhancement unit, an environment feature sampling unit, and a context-based adaptive fusion update unit. Outputs are analyzed with a graph neural network and constructed into a topological representation of essential features, resulting in a target category vector. The t-Distributed Stochastic Neighbor Embedding (t-SNE) method is used to qualitatively evaluate the algorithm’s classification ability, while metrics like accuracy, recall, and F1 score are used to quantitatively analyze key units and overall network performance. Additionally, class activation map visualization methods are employed to validate the extraction and analysis of invariant features at different stages and branches. The proposed method achieves recognition accuracies of 98.56% on the MSTAR dataset, 94.11% on SAR-ACD dataset, and 86.20% on OpenSARShip dataset, demonstrating its effectiveness in extracting essential target features.
format Article
id doaj-art-55a6f64b154c4f949991d7ec2d9cbf86
institution DOAJ
issn 2095-283X
language English
publishDate 2025-04-01
publisher China Science Publishing & Media Ltd. (CSPM)
record_format Article
series Leida xuebao
spelling doaj-art-55a6f64b154c4f949991d7ec2d9cbf862025-08-20T02:53:33ZengChina Science Publishing & Media Ltd. (CSPM)Leida xuebao2095-283X2025-04-0114236638810.12000/JR24125R24125Target Recognition Method Based on Graph Structure Perception of Invariant Features for SAR ImagesJingyi CAO0Yang ZHANG1Ya’nan YOU2Yamin WANG3Feng YANG4Weijia REN5Jun LIU6School of Artificial Intelligence, Beijing University of Posts and Telecommunications, Beijing 100876, ChinaSchool of Artificial Intelligence, Beijing University of Posts and Telecommunications, Beijing 100876, ChinaSchool of Artificial Intelligence, Beijing University of Posts and Telecommunications, Beijing 100876, ChinaSchool of Electronic and Information Engineering, Beihang University, Beijing 100191, ChinaChangsha Tianyi Space Science and Technology Research Institute Co., Ltd., Changsha 410221, ChinaChangsha Tianyi Space Science and Technology Research Institute Co., Ltd., Changsha 410221, ChinaSchool of Artificial Intelligence, Beijing University of Posts and Telecommunications, Beijing 100876, ChinaSynthetic Aperture Radar (SAR) image target recognition technology based on deep learning has matured. However, challenges remain due to scattering phenomenon and noise interference that cause significant intraclass variability in imaging results. Invariant features, which represent the essential attributes of a specific target class with consistent expressions, are crucial for high-precision recognition. We define these invariant features from the entity, its surrounding environment, and their combined context as the target’s essential features. Guided by multilevel essential feature modeling theory, we propose a SAR image target recognition method based on graph networks and invariant feature perception. This method employs a dual-branch network to process multiview SAR images simultaneously using a rotation-learnable unit to adaptively align dual-branch features and reinforce invariant features with rotational immunity by minimizing intraclass feature differences. Specifically, to support essential feature extraction in each branch, we design a feature-guided graph feature perception module based on multilevel essential feature modeling. This module uses salient points for target feature analysis and comprises a target ontology feature enhancement unit, an environment feature sampling unit, and a context-based adaptive fusion update unit. Outputs are analyzed with a graph neural network and constructed into a topological representation of essential features, resulting in a target category vector. The t-Distributed Stochastic Neighbor Embedding (t-SNE) method is used to qualitatively evaluate the algorithm’s classification ability, while metrics like accuracy, recall, and F1 score are used to quantitatively analyze key units and overall network performance. Additionally, class activation map visualization methods are employed to validate the extraction and analysis of invariant features at different stages and branches. The proposed method achieves recognition accuracies of 98.56% on the MSTAR dataset, 94.11% on SAR-ACD dataset, and 86.20% on OpenSARShip dataset, demonstrating its effectiveness in extracting essential target features.https://radars.ac.cn/cn/article/doi/10.12000/JR24125synthetic aperture radar (sar)target recognitioninvariant feature extractionessential featuredeep learning
spellingShingle Jingyi CAO
Yang ZHANG
Ya’nan YOU
Yamin WANG
Feng YANG
Weijia REN
Jun LIU
Target Recognition Method Based on Graph Structure Perception of Invariant Features for SAR Images
Leida xuebao
synthetic aperture radar (sar)
target recognition
invariant feature extraction
essential feature
deep learning
title Target Recognition Method Based on Graph Structure Perception of Invariant Features for SAR Images
title_full Target Recognition Method Based on Graph Structure Perception of Invariant Features for SAR Images
title_fullStr Target Recognition Method Based on Graph Structure Perception of Invariant Features for SAR Images
title_full_unstemmed Target Recognition Method Based on Graph Structure Perception of Invariant Features for SAR Images
title_short Target Recognition Method Based on Graph Structure Perception of Invariant Features for SAR Images
title_sort target recognition method based on graph structure perception of invariant features for sar images
topic synthetic aperture radar (sar)
target recognition
invariant feature extraction
essential feature
deep learning
url https://radars.ac.cn/cn/article/doi/10.12000/JR24125
work_keys_str_mv AT jingyicao targetrecognitionmethodbasedongraphstructureperceptionofinvariantfeaturesforsarimages
AT yangzhang targetrecognitionmethodbasedongraphstructureperceptionofinvariantfeaturesforsarimages
AT yananyou targetrecognitionmethodbasedongraphstructureperceptionofinvariantfeaturesforsarimages
AT yaminwang targetrecognitionmethodbasedongraphstructureperceptionofinvariantfeaturesforsarimages
AT fengyang targetrecognitionmethodbasedongraphstructureperceptionofinvariantfeaturesforsarimages
AT weijiaren targetrecognitionmethodbasedongraphstructureperceptionofinvariantfeaturesforsarimages
AT junliu targetrecognitionmethodbasedongraphstructureperceptionofinvariantfeaturesforsarimages