Structurization of synthetic aperture radar information by using neural networks

Recent earth observation data gathered by satellite-borne synthetic aperture radar (SAR) systems grow vei7 rapidly. Previously we proposed the structurization of the gathered data into an easy-to-use form for extensive utilization of the data. There we employ convolutional neural networks to extract lands shape features to realize automatic metrization of local area patches based on teiTain features obtained as SAR images. This paper reviews successful metrization in the proposal, which leads to total big-SAR-data structurization.