Winner Takes All: A Superpixel Aided Voting Algorithm for Training Unsupervised PolSAR CNN Classifiers

Unsupervised methods play an essential role in polarimetric synthetic aperture radar (PolSAR) image classification, where labeled data are difficult to obtain. However, there is still a large gap between existing unsupervised learning methods and supervised learning methods. Without the semantic con...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on geoscience and remote sensing 2022, Vol.60, p.1-19
Main Authors: Zuo, Yixin, Guo, Jiayi, Zhang, Yueting, Hu, Yuxin, Lei, Bin, Qiu, Xiaolan, Ding, Chibiao
Format: Article
Language:English
Subjects:
Accuracy
Airborne radar
Airborne remote sensing
Algorithms
Artificial neural networks
Convolutional neural network (CNN) model
Feature extraction
Image classification
Methods
multiscale (MS) semantic feature
Neural networks
Pixels
polarimetric synthetic aperture radar (PolSAR) image
Radar
Radar imaging
SAR (radar)
Scattering
Semantics
Submarine pipelines
superpixel segmentation
Supervised learning
Synthetic aperture radar
Teaching methods
Training
unsupervised classification
Unsupervised learning
Voting
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Unsupervised methods play an essential role in polarimetric synthetic aperture radar (PolSAR) image classification, where labeled data are difficult to obtain. However, there is still a large gap between existing unsupervised learning methods and supervised learning methods. Without the semantic constraints of labeled data, pixels within the same category are often misclassified into different categories, leaving the output to be messy. To address the previous issue, this article proposes a fully unsupervised pipeline for training convolutional neural networks (CNNs). The pipeline combines low-level superpixels and high-level CNN semantic features for high-quality pseudolabel generation. It effectively eliminates the misclassified pixels by voting within the superpixel blob while preserving the sharpness of edges. With the training process of the model, the quality of the generated labels is getting improved. Experiments on airborne [experimental airborne SAR system from Germany (ESAR)/airborne synthetic aperture radar from America (AIRSAR)] and spaceborne (RadarSat2) PolSAR images prove the effectiveness of the proposed method (measured with overall accuracy (OA), average accuracy (AA), and Kappa metrics). Our method outperforms the previous unsupervised methods (H/alpha-Wishart, SM-Wishart, FDD-H, DEC, and VQC-CAE) with a large margin and even has comparable performance to the supervised CNN model [fully CNN (FCN)].
ISSN:0196-2892
1558-0644
DOI:10.1109/TGRS.2022.3177900