EXPLORING GROUND SEGMENTATION FROM LIDAR SCANNING-DERIVED IMAGES USING CONVOLUTIONAL NEURAL NETWORKS

The method for capturing two-dimensional surface displacements was applied to three pairs of pre- and post-event high-resolution TerraSAR-X intensity images of the 2011 Tohoku-Oki earthquake taken in ascending and descending paths, and the results were used for estimating the actual three-dimensional movements. Compared with the GPS data at three ground-control stations, the estimation errors in the horizontal and vertical directions were less than 0.3 m in spite of the pixels size (1.25 m).