SAR-based waterbody detection using morphological feature extraction and integration

Abstract

In this paper, the authors have analysed the signature of extracting water body from SAR intensity and coherence images with morphological segmentation, and presented a solution to water body mapping in the absence of ancillary information from air or ground. The use of morphological image reconstruction algorithms is a great improvement for mapping water with SAR imagery. Especially, the innovative decision fusion method on integrating intermediate extracted data from intensity and coherence images is the first of its kind in water body mapping. The image reconstruction from fusion marker and mask has shown its promise in removing non-water related features/patterns. A case study is given to demonstrate the efficiency and accuracy of this proposed methodology in removing vegetation cover and man-made features. Thus, the advantage of SAR imagery available unlimited by weather condition can be fully exploited. Furthermore, due to the reduced computing time, this approach for water body mapping can assist authorities to respond promptly to emergencies such as major floods.