Dual-Branch Fusion of Convolutional Neural Network and Graph Convolutional Network for PolSAR Image Classification
Polarimetric synthetic aperture radar (PolSAR) images contain useful information, which can lead to extensive land cover interpretation and a variety of output products. In contrast to optical imagery, there are several challenges in extracting beneficial features from PolSAR data. Deep learning (DL...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-12-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/1/75 |
| _version_ | 1827597200145776640 |
|---|---|
| author | Ali Radman Masoud Mahdianpari Brian Brisco Bahram Salehi Fariba Mohammadimanesh |
| author_facet | Ali Radman Masoud Mahdianpari Brian Brisco Bahram Salehi Fariba Mohammadimanesh |
| author_sort | Ali Radman |
| collection | DOAJ |
| description | Polarimetric synthetic aperture radar (PolSAR) images contain useful information, which can lead to extensive land cover interpretation and a variety of output products. In contrast to optical imagery, there are several challenges in extracting beneficial features from PolSAR data. Deep learning (DL) methods can provide solutions to address PolSAR feature extraction challenges. The convolutional neural networks (CNNs) and graph convolutional networks (GCNs) can drive PolSAR image characteristics by deploying kernel abilities in considering neighborhood (local) information and graphs in considering long-range similarities. A novel dual-branch fusion of CNN and mini-GCN is proposed in this study for PolSAR image classification. To fully utilize the PolSAR image capacity, different spatial-based and polarimetric-based features are incorporated into CNN and mini-GCN branches of the proposed model. The performance of the proposed method is verified by comparing the classification results to multiple state-of-the-art approaches on the airborne synthetic aperture radar (AIRSAR) dataset of Flevoland and San Francisco. The proposed approach showed 1.3% and 2.7% improvements in overall accuracy compared to conventional methods with these AIRSAR datasets. Meanwhile, it enhanced its one-branch version by 0.73% and 1.82%. Analyses over Flevoland data further indicated the effectiveness of the dual-branch model using varied training sampling ratios, leading to a promising overall accuracy of 99.9% with a 10% sampling ratio. |
| first_indexed | 2024-03-09T03:26:04Z |
| format | Article |
| id | doaj.art-11645fe423fc491583006f322da455d1 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-09T03:26:04Z |
| publishDate | 2022-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-11645fe423fc491583006f322da455d12023-12-03T15:02:11ZengMDPI AGRemote Sensing2072-42922022-12-011517510.3390/rs15010075Dual-Branch Fusion of Convolutional Neural Network and Graph Convolutional Network for PolSAR Image ClassificationAli Radman0Masoud Mahdianpari1Brian Brisco2Bahram Salehi3Fariba Mohammadimanesh4Department of Electrical and Computer Engineering, Memorial University of Newfoundland, St. John’s, NL A1B3X5, CanadaDepartment of Electrical and Computer Engineering, Memorial University of Newfoundland, St. John’s, NL A1B3X5, CanadaThe Canada Centre for Mapping and Earth Observation, Ottawa, ON K1S 5K2, CanadaDepartment of Environmental Resources Engineering, State University of New York College of Environmental Science and Forestry (SUNY ESF), Syracuse, NY 13210, USAC-CORE, St. John’s, NL A1B 3X5, CanadaPolarimetric synthetic aperture radar (PolSAR) images contain useful information, which can lead to extensive land cover interpretation and a variety of output products. In contrast to optical imagery, there are several challenges in extracting beneficial features from PolSAR data. Deep learning (DL) methods can provide solutions to address PolSAR feature extraction challenges. The convolutional neural networks (CNNs) and graph convolutional networks (GCNs) can drive PolSAR image characteristics by deploying kernel abilities in considering neighborhood (local) information and graphs in considering long-range similarities. A novel dual-branch fusion of CNN and mini-GCN is proposed in this study for PolSAR image classification. To fully utilize the PolSAR image capacity, different spatial-based and polarimetric-based features are incorporated into CNN and mini-GCN branches of the proposed model. The performance of the proposed method is verified by comparing the classification results to multiple state-of-the-art approaches on the airborne synthetic aperture radar (AIRSAR) dataset of Flevoland and San Francisco. The proposed approach showed 1.3% and 2.7% improvements in overall accuracy compared to conventional methods with these AIRSAR datasets. Meanwhile, it enhanced its one-branch version by 0.73% and 1.82%. Analyses over Flevoland data further indicated the effectiveness of the dual-branch model using varied training sampling ratios, leading to a promising overall accuracy of 99.9% with a 10% sampling ratio.https://www.mdpi.com/2072-4292/15/1/75classificationconvolutional neural network (CNNs)dual-branch fusiongraph convolutional networks (GCNs)PolSAR |
| spellingShingle | Ali Radman Masoud Mahdianpari Brian Brisco Bahram Salehi Fariba Mohammadimanesh Dual-Branch Fusion of Convolutional Neural Network and Graph Convolutional Network for PolSAR Image Classification Remote Sensing classification convolutional neural network (CNNs) dual-branch fusion graph convolutional networks (GCNs) PolSAR |
| title | Dual-Branch Fusion of Convolutional Neural Network and Graph Convolutional Network for PolSAR Image Classification |
| title_full | Dual-Branch Fusion of Convolutional Neural Network and Graph Convolutional Network for PolSAR Image Classification |
| title_fullStr | Dual-Branch Fusion of Convolutional Neural Network and Graph Convolutional Network for PolSAR Image Classification |
| title_full_unstemmed | Dual-Branch Fusion of Convolutional Neural Network and Graph Convolutional Network for PolSAR Image Classification |
| title_short | Dual-Branch Fusion of Convolutional Neural Network and Graph Convolutional Network for PolSAR Image Classification |
| title_sort | dual branch fusion of convolutional neural network and graph convolutional network for polsar image classification |
| topic | classification convolutional neural network (CNNs) dual-branch fusion graph convolutional networks (GCNs) PolSAR |
| url | https://www.mdpi.com/2072-4292/15/1/75 |
| work_keys_str_mv | AT aliradman dualbranchfusionofconvolutionalneuralnetworkandgraphconvolutionalnetworkforpolsarimageclassification AT masoudmahdianpari dualbranchfusionofconvolutionalneuralnetworkandgraphconvolutionalnetworkforpolsarimageclassification AT brianbrisco dualbranchfusionofconvolutionalneuralnetworkandgraphconvolutionalnetworkforpolsarimageclassification AT bahramsalehi dualbranchfusionofconvolutionalneuralnetworkandgraphconvolutionalnetworkforpolsarimageclassification AT faribamohammadimanesh dualbranchfusionofconvolutionalneuralnetworkandgraphconvolutionalnetworkforpolsarimageclassification |