Disaster Area Detection from Synthetic Aperture Radar Images Using Convolutional Autoencoder and One-class SVM

Authors
Shingo Mabu*, Kohki Fujita, Takashi Kuremoto
Graduate School of Sciences and Technology for Innovation, Yamaguchi University, Tokiwadai 2-16-1, Ube, Yamaguchi 755-8611, Japan
*Corresponding author. Email: [email protected]
Corresponding Author
Shingo Mabu
Received 15 November 2017, Accepted 18 December 2017, Available Online 25 June 2019.
DOI
https://doi.org/10.2991/jrnal.k.190601.001
Keywords
Anomaly detection; convolutional autoencoder; one-class SVM; synthetic aperture radar
Abstract
In recent years, research on detecting disaster areas from synthetic aperture radar (SAR) images has been conducted. When machine learning is used for disaster area detection, a large number of training data are required; however, we cannot obtain so much training data with correct class labels. Therefore, in this research, we propose an anomaly detection system that finds abnormal areas that deviate from normal ones. The proposed method uses a convolutional autoencoder (CAE) for feature extraction and one-class support vector machine (OCSVM) for anomaly detection.
Copyright
© 2019 The Authors. Published by ALife Robotics Corp. Ltd.
Open Access
This is an open access article distributed under the CC BY-NC 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)