Learning-based detection and time-series prediction of ground motion InSAR data
Dr Pui Anantrasirichai and Dr Paul Hill, Research Fellows within the Visual Information Laboratory, University of Bristol
Recent improvements in the frequency, type and availability of satellite images mean it is now feasible to forecast ground motion in general and specifically monitor volcanoes in remote and inaccessible regions.
This talk demonstrates the capability of machine learning algorithms for both detecting volcanic ground deformation and ground motion prediction in general within large sets of Interferometric Synthetic Aperture Radar (InSAR) data. Proposed detection and prediction methods include the use of Convolutional Neural Networks (CNN), SARIMA and LSTMs. Volcanic unrest together with small, slow rate of ground movement was effectively detected using CNNs. Although challenging to forecast, InSAR time series were able to be effectively predicted for different types of signals using SARIMA and LSTM methods.
Both Pui and Paul have worked on a range of projects combining remote sensing and machine learning and are excited to be given the opportunity to present results from recent studies to the wider vision community.
Dr Pui Anantrasirichai, Research Fellow, Visual Information Laboratory, University of Bristol
Pui received her B.E. degree in Electrical Engineering from Chulalongkorn University, Thailand, in 2000, and her Ph.D. degree in Electrical and Electronic Eengineering from the University of Bristol, in 2007. She is currently a Senior Research Fellow with the Visual Information Laboratory, University of Bristol. Her current research interests include image and video coding, image analysis and enhancement, medical imaging, texture-based image analysis, machine learning and remote sensing. She has been involved in many projects across multiple disciplines, including telecommunications, biology, optometry, clinical sciences, biochemistry, mechanical engineering, psychology and geoscience.
Dr Paul Hill, Senior Research Fellow in Image Processing, Visual Information Laboratory, University of Bristol
Paul received his B.Sc degree from the Open University (1996), and M.Sc degree from the University of Bristol, (1998), and a Ph.D. also from the University of Bristol (2002). His research interests include image and video analysis, compression, fusion, and multiscale transforms together with audio applications such as compression, retrieval, and signal separation. He is currently a Senior Research Fellow at the Department of Electrical and Electronic Engineering at the University of Bristol. His current research directions have included many different combinations of remote sensing and machine learning technologies including volcano, and Harmful Algal Bloom (HAB) detection methods. He has recently completed a book on signal processing using Matlab with CRC press.
For anyone who missed the seminar, it is available to watch here.