A system for automated vision-based sea-ice concentration detection and floe-size distribution indication from an icebreaker

Abstract

This paper presents a ship-mounted multi-lens camera system for sea-ice monitoring and algorithms to automatically evaluate the sea-ice concentration and to indicate the floe-sizes in a radius of 100 meter around the vessel. During the SWEDARCTIC Arctic Ocean 2016 expedition, 11 camera lenses recorded the sea-ice conditions around the Swedish icebreaker Oden. As an example of the possible use of this image system, the images of six lenses are combined into one 360° panoramic image. To distinguish between water and sea-ice in the images, and thus to evaluate the sea-ice concentration around the vessel, a direct thresholding, the k-means, and a novel adaptive thresholding method are applied. Moreover, an edge detector gives the number of pixels that either form the boundary between sea-ice and water or are part of a visible ice fracture. The ratio between these edge pixels and the total number of pixels containing sea-ice gives an indication of the floe size distribution (FSD) in the image.