Multi-Dimensional Discrete Wavelet Transformation and Its Application to Polarimetric SAR Classification
No ratings
Presented at SAIcompconf 2016 by
Professor, Saga University & Adjunct Professor, University of Arizona Dr Kohei Arai, a Scientist, Professor and Author. He is currently Professor at Saga University, Japan and Adjunct Prof. of the University of Arizona, USA since 1998. Dr Arai received PhD degree in Information Science from Nihon University in June 1982 and MS degree in Electronics Engineering from Nihon University in March 1974. His current research concerns are Satellite Remote Sensing, Radiative Transfer Equation, Human-Computer Interaction, Image Recognition and Understanding, Non-Linear Optimization Theory and Wavelet Analysis. Dr Arai holds 42 patents and received numerous awards, including the Patent Award of the Year. Dr Arai has been featured in Japan Times and Italian Newspapers for his work on Eyes only Computer System. He has worked on several global research collaboration projects during his career. He wrote 31 books and published 490 journal papers and 390 of conference papers. Abstract: Discrete Wavelet Transformation: DWT is explained in a comprehensive manner starting from one dimensional to multi-dimensional DWT. One of the applications of the multi-dimensional DWT is introduced for polarimetric Synthetic Aperture Radar: SAR classification of sea ice type discriminations. Polarimetric SAR onboard remote sensing satellite allows sea ice monitoring as well as land use classifications. One of the specific features of polarimetric SAR is to clarify scattering properties of the ground cover targets. By using the differences of the scattering properties, land use classifications as well as sea ice discriminations can be done. In order to extract scattering properties, scattering matrix is derived from the polarimetric SAR data. From the scattering matrix, trajectory in a phase space is calculated. On the other hand, time and frequency analysis is available by the multi-dimensional DWT. In order to use the specific feature of the trajectory in the phase space, behavior of the scattering properties, multi-dimensional DWT is applied to the trajectory. Thus the specific polarimetric features of the designated ground cover targets are extracted and classified using supervised classifications.