| Title | Automated method for SAR-based water-mask derivation |
| Download | Downloads |
| |
| Licence | Please note the adoption of the Open Government Licence - Canada
supersedes any previous licences. |
| Author | Beauchemin, M; Decker, V; Tolszczuk-Leclerc, S |
| Source | Geomatics Canada, Open File 35, 2017, 27 pages, https://doi.org/10.4095/304204  Open Access |
| Year | 2017 |
| Publisher | Natural Resources Canada |
| Document | open file |
| Lang. | English |
| Media | on-line; digital |
| File format | pdf |
| Subjects | geophysics; remote sensing; satellite imagery; radar methods; radar imagery; surface waters; floods; vegetation; mapping techniques |
| Illustrations | flow charts; tables; satellite images |
| Program | Remote Sensing Science |
| Released | 2017 07 17 |
| Abstract | We describe an automated method for water mask extraction from RADARSAT images. The method is specifically designed to localize image areas where conditions are best for the optimum application of
automated thresholding algorithms. The approach is rooted in a split-based technique, successfully exploited elsewhere, to derive a global thresholding value from the selection of a few local tiles. The method further incorporates a region growing
mechanism to locally improve the waterline boundary position. It also includes an adaptive algorithm for small land patches conditional reclassification. The algorithm was applied to 18 RADARSAT images and careful visual inspection of the results
shows good agreement with results obtained manually by experienced operators using a traditional threshold approach. |
| Summary | (Plain Language Summary, not published)
We describe a computer method for automated water boundary delineation from radar images obtained by the Canadian RADARSAT satellite. Because water in
radar imagery typically appears as darker areas, the method is based on algorithms that separate bright areas from dark ones. The bright-dark transition locations define the water boundaries. The method was applied on 18 RADARSAT images, and careful
visual inspection of the results shows good agreement with results obtained manually by experienced operators. The method was developed in support of the existing manual water delineation workflow currently employed by the Emergency Geomatics Service
(EGS) at the Canada Centre for Mapping and Earth Observation (CCMEO). |
| GEOSCAN ID | 304204 |
|
|