Investigation into Techniques for Real-Time Crowded Ice Floe Segmentation

The following paper describes recent efforts made by the National Research Council of Canada (NRC) in investigating improved approaches to image segmentation of managed ice floes, particularly in physical-scale model testing basins. A segmentation approach previously implemented by NRC is compared against a method utilizing Gradient Vector Flow (GVF) Snakes, and a modified version of this approach with potentially improved run-time performance is discussed.