Automatic Channel Network Extraction From Remotely Sensed Images by Singularity Analysis

Abstract

Quantitative analysis of channel networks plays an important role in river\nstudies. To provide a quantitative representation of channel networks, we\npropose a new method that extracts channels from remotely sensed images and\nestimates their widths. Our fully automated method is based on a recently\nproposed Multiscale Singularity Index that responds strongly to curvilinear\nstructures but weakly to edges. The algorithm produces a channel map, using a\nsingle image where water and non-water pixels have contrast, such as a Landsat\nnear-infrared band image or a water index defined on multiple bands. The\nproposed method provides a robust alternative to the procedures that are used\nin remote sensing of fluvial geomorphology and makes classification and\nanalysis of channel networks easier. The source code of the algorithm is\navailable at: http://live.ece.utexas.edu/research/cne/.\n