Multi-Scale Feature Extraction with 3D Complex-Valued Network for PolSAR Image Classification

Compared to traditional real-valued neural networks, which process only amplitude information, complex-valued neural networks handle both amplitude and phase information, leading to superior performance in polarimetric synthetic aperture radar (PolSAR) image classification tasks. This paper proposes...

Full description

Saved in:
Bibliographic Details
Main Authors: Nana Jiang, Wenbo Zhao, Jiao Guo, Qiang Zhao, Jubo Zhu
Format: Article
Language:English
Published: MDPI AG 2025-08-01
Series:Remote Sensing
Subjects:
Online Access:https://www.mdpi.com/2072-4292/17/15/2663
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849406446999961600
author Nana Jiang
Wenbo Zhao
Jiao Guo
Qiang Zhao
Jubo Zhu
author_facet Nana Jiang
Wenbo Zhao
Jiao Guo
Qiang Zhao
Jubo Zhu
author_sort Nana Jiang
collection DOAJ
description Compared to traditional real-valued neural networks, which process only amplitude information, complex-valued neural networks handle both amplitude and phase information, leading to superior performance in polarimetric synthetic aperture radar (PolSAR) image classification tasks. This paper proposes a multi-scale feature extraction (MSFE) method based on a 3D complex-valued network to improve classification accuracy by fully leveraging multi-scale features, including phase information. We first designed a complex-valued three-dimensional network framework combining complex-valued 3D convolution (CV-3DConv) with complex-valued squeeze-and-excitation (CV-SE) modules. This framework is capable of simultaneously capturing spatial and polarimetric features, including both amplitude and phase information, from PolSAR images. Furthermore, to address robustness degradation from limited labeled samples, we introduced a multi-scale learning strategy that jointly models global and local features. Specifically, global features extract overall semantic information, while local features help the network capture region-specific semantics. This strategy enhances information utilization by integrating multi-scale receptive fields, complementing feature advantages. Extensive experiments on four benchmark datasets demonstrated that the proposed method outperforms various comparison methods, maintaining high classification accuracy across different sampling rates, thus validating its effectiveness and robustness.
format Article
id doaj-art-ba916ee702a845c1a3afaec486900704
institution Kabale University
issn 2072-4292
language English
publishDate 2025-08-01
publisher MDPI AG
record_format Article
series Remote Sensing
spelling doaj-art-ba916ee702a845c1a3afaec4869007042025-08-20T03:36:22ZengMDPI AGRemote Sensing2072-42922025-08-011715266310.3390/rs17152663Multi-Scale Feature Extraction with 3D Complex-Valued Network for PolSAR Image ClassificationNana Jiang0Wenbo Zhao1Jiao Guo2Qiang Zhao3Jubo Zhu4School of Artificial Intelligence, Sun Yat-sen University, Zhuhai 519082, ChinaSchool of Artificial Intelligence, Sun Yat-sen University, Zhuhai 519082, ChinaCollege of Mechanical and Electronic Engineering, Northwest A&F University, Yangling 712100, ChinaShanghai Institute of Satellite Engineering, Shanghai 201109, ChinaSchool of Artificial Intelligence, Sun Yat-sen University, Zhuhai 519082, ChinaCompared to traditional real-valued neural networks, which process only amplitude information, complex-valued neural networks handle both amplitude and phase information, leading to superior performance in polarimetric synthetic aperture radar (PolSAR) image classification tasks. This paper proposes a multi-scale feature extraction (MSFE) method based on a 3D complex-valued network to improve classification accuracy by fully leveraging multi-scale features, including phase information. We first designed a complex-valued three-dimensional network framework combining complex-valued 3D convolution (CV-3DConv) with complex-valued squeeze-and-excitation (CV-SE) modules. This framework is capable of simultaneously capturing spatial and polarimetric features, including both amplitude and phase information, from PolSAR images. Furthermore, to address robustness degradation from limited labeled samples, we introduced a multi-scale learning strategy that jointly models global and local features. Specifically, global features extract overall semantic information, while local features help the network capture region-specific semantics. This strategy enhances information utilization by integrating multi-scale receptive fields, complementing feature advantages. Extensive experiments on four benchmark datasets demonstrated that the proposed method outperforms various comparison methods, maintaining high classification accuracy across different sampling rates, thus validating its effectiveness and robustness.https://www.mdpi.com/2072-4292/17/15/2663complex-valued networkcomplex-valued 3D convolutionmulti-scale featurepolarimetric synthetic aperture radar (PolSAR) image classification
spellingShingle Nana Jiang
Wenbo Zhao
Jiao Guo
Qiang Zhao
Jubo Zhu
Multi-Scale Feature Extraction with 3D Complex-Valued Network for PolSAR Image Classification
Remote Sensing
complex-valued network
complex-valued 3D convolution
multi-scale feature
polarimetric synthetic aperture radar (PolSAR) image classification
title Multi-Scale Feature Extraction with 3D Complex-Valued Network for PolSAR Image Classification
title_full Multi-Scale Feature Extraction with 3D Complex-Valued Network for PolSAR Image Classification
title_fullStr Multi-Scale Feature Extraction with 3D Complex-Valued Network for PolSAR Image Classification
title_full_unstemmed Multi-Scale Feature Extraction with 3D Complex-Valued Network for PolSAR Image Classification
title_short Multi-Scale Feature Extraction with 3D Complex-Valued Network for PolSAR Image Classification
title_sort multi scale feature extraction with 3d complex valued network for polsar image classification
topic complex-valued network
complex-valued 3D convolution
multi-scale feature
polarimetric synthetic aperture radar (PolSAR) image classification
url https://www.mdpi.com/2072-4292/17/15/2663
work_keys_str_mv AT nanajiang multiscalefeatureextractionwith3dcomplexvaluednetworkforpolsarimageclassification
AT wenbozhao multiscalefeatureextractionwith3dcomplexvaluednetworkforpolsarimageclassification
AT jiaoguo multiscalefeatureextractionwith3dcomplexvaluednetworkforpolsarimageclassification
AT qiangzhao multiscalefeatureextractionwith3dcomplexvaluednetworkforpolsarimageclassification
AT jubozhu multiscalefeatureextractionwith3dcomplexvaluednetworkforpolsarimageclassification