Proposal of Complex-Valued Convolutional Neural Networks for Similar Land-Shape Discovery in Interferometric Synthetic Aperture Radar
We propose a complex-valued convolutional neural network to extract the areas having land shapes similar to samples in interferometric synthetic aperture radar (InSAR). InSAR extends its application to various earth observations such as volcano monitoring and earthquake damage estimation. Since the amount of data is increasing drastically in these years, it is necessary to structurize them in a big data framework. In this paper, experiments demonstrate that similar small volcanoes are grouped into a single class. We find that the neural network is capable of discovering unidentified lands similar to prepared samples successfully.
KeywordsInterferometric synthetic aperture radar (InSAR) Feature discovery Complex-valued neural network (CVNN)
A part of this work was supported by JSPS KAKENHI Grant Numbers 15H02756 and 18H04105, and also by the Cooperative Research Project Program of the Research Institute of Electrical Communication (RIEC), Tohoku University. The Advanced Land Observing Satellite (ALOS) original data are copyrighted by Japan Aerospace Exploration Agency (JAXA) and provided under JAXA Fourth ALOS Research Announcement PI No. 1154.
- 1.G-portal. https://www.gportal.jaxa.jp/gp/top.html
- 7.Hirose, A.: Applications of complex-valued neural networks to coherent optical computing using phase-sensitive detection scheme. Inf. Sci.-Appl. 2, 103–117 (1994)Google Scholar
- 10.Hirose, A., Tsuda, S., Natsuaki, R.: Structurization of synthetic aperture radar information by using neural networks. In: SAR in Big Data Era: Models, Methods and Applications (BIGSARDATA) 2017, pp. 1–4, November 2017. https://doi.org/10.1109/BIGSARDATA.2017.8124936
- 12.Kim, H., Hirose, A.: Codebook-based hierarchical polarization feature for unsupervised fine land classification using high-resolution PolSAR data. In: International Geoscience and Remote Sensing Symposium (IGARSS) 2018 Valencia. IEEE (to be presented 2018)Google Scholar
- 13.Kim, H., Hirose, A.: Polarization feature extraction using quaternion neural networks for flexible unsupervised PolSAR land classification. In: International Geoscience and Remote Sensing Symposium (IGARSS) 2018 Valencia, to be presented, July 2018Google Scholar
- 19.Suksmono, A.B., Hirose, A.: Adaptive complex-amplitude texture classifier that deals with both height and reflectance for interferometric SAR images. IEICE Trans. Electron. E83–C(12), 1912–1916 (2000)Google Scholar