University of Waterloo

Addressing Non-stationarity in SAR Imagery for Sea Ice Segmentation

Max Manning, MASc candidate

Mar 25, 2022, 11:30am EC4-2101A

The backscatter signature of sea ice observed in synthetic aperture radar imagery is modulated by a number of external factors including sensor geometry and environmental conditions. Segmenting sea ice in SAR scenes is therefore a non-stationary classification problem. I will be discussing some of the complicating factors that make sea ice segmentation difficult, and presenting some of my recent work on addressing these challenges with spatially varying parameter models based on kernel estimation.