Navigation Assistance for Ice-Infested Waters Through Automatic Iceberg Detection and Ice Classification Based on Terrasar-X Imagery

Abstract

ABSTRACT:Most icebergs present in northern latitudes originate from western Greenland glaciers, from where they drift into Baffin Bay, circulating north along Greenland coast and south along Canadian coast. Some of them drift more southwards up to Newfoundland, where they frequently cross shipping routes. Furthermore, the Arctic summer sea ice coverage significantly decreased over the last three decades. This has attracted numerous attentions from maritime end-users. To keep Arctic shipping routes safe, the monitoring of sea ice and icebergs is crucial. For this purpose, satellite-based Synthetic Aperture Radar (SAR) is well suited. Equipped with an active radar antenna, SAR satellites provide image data of the ocean and frozen waters independent of weather conditions, cloud cover or absence of daylight. In this paper, we present a processor for sea ice classification and (subsequent) iceberg detection based on TerraSAR-X imagery. In the classification step, texture features are extracted from the images and fed into a neural network, indicating areas of low sea ice concentration. Then, an adapted Constant False Alarm Rate (CFAR) detector is executed in order to detect icebergs. In the end, sea ice boundary and iceberg positions are output. Our experiments deal with HH polarized TerraSAR-X images taken in spring season in the Baffin Bay off the western Greenland coast, where both, sea ice and icebergs are present. Our results exemplify how a comprehensive ice processor with complementary information can be set up for near real time (NRT) service in ice infested waters.

Automated Iceberg Detection Using High-Resolution X-Band SAR Images

“…Gao et al (2009) introduced an iterative concept of the CFAR detector for terrestrial target detection in dense traffic situations, however, thus far it has not been applied to iceberg detection. In our previous work, we adapted this iterative approach and applied it to TerraSAR-X ScanSAR images containing a high iceberg density (Ressel et al 2015). …”

“…In open water, low-gradient values predominate. In former studies, e.g., Lehner et al (2014) and Ressel et al (2015), gradient-related features are used to classify different ice types. Here, the binarized gradient is used to initialize the mask.…”

“…For detection, we make use of the constant false alarm rate (CFAR) detector, applying the iterative concept as mentioned by Ressel et al (2015). Building on our previous work and on former studies in wind and sea state retrieval, we survey patterns in SAR images (i.e., waves) and integrate a new wave filter into our automated methodology.…”

Investigation into Different Polarimetric Features for Sea Ice Classification Using X-Band Synthetic Aperture Radar

“…This facilitates wide area investigations, e.g., as conducted for field experiments in the marginal ice zone (ONR MIZ 2015 campaign, see [1]). Topics in sea ice research using SAR are ice drift (see [2]- [6]), sea state and wave propagation into sea ice (see [7] and [8]), ice concentration (see [9]), iceberg detection (see [10] and [11]). Most research published so far on SAR-based sea ice classification concentrates on single polarized data (e.g., [12]- [22]).…”

The Neural-Fuzzy Approach as a Way of Preventing a Maritime Vessel Accident in a Heavy Traffic Zone

“…For example, the paper [16] focuses on the neural network classification of marine targets on noise images in bad weather conditions and in case of sea waves. In the paper [17] the neural network succeeds in detecting icebergs by typical textural features from the image of satellite-based Synthetic Aperture Radar. Moreover, the neural network has been used successfully in issues of improving an accurate position of a vessel from radar data [18] and designing laser metering systems [19].…”