Automatic detection of flooded areas on ENVISAT/ASAR images using an object-oriented classification technique and an active contour algorithm

Abstract

Two techniques for extracting flooded areas out of radar-imagery in a time-efficient way are presented in this paper. The results of the object-oriented classification technique, based on the commercial eCognition software, and the active contour technique, are both obtained on 2 ENVISAT ASAR images; one recorded during a flood period in Flanders on 2 January 2003 and one during a non-flooded region on 26 June of the same year. In both techniques the net flooded result is based on the subtraction of the existing water bodies (i.e. lakes, rivers, canals, ...) obtained from the non-flooded reference image, from the image recorded during the flooded period.