Picture water droplets

Developing mathematical theories of the physical world: Open Access research on fluid dynamics from Strathclyde

Strathprints makes available Open Access scholarly outputs by Strathclyde's Department of Mathematics & Statistics, where continuum mechanics and industrial mathematics is a specialism. Such research seeks to understand fluid dynamics, among many other related areas such as liquid crystals and droplet evaporation.

The Department of Mathematics & Statistics also demonstrates expertise in population modelling & epidemiology, stochastic analysis, applied analysis and scientific computing. Access world leading mathematical and statistical Open Access research!

Explore all Strathclyde Open Access research...

Effective SAR image segmentation and sea-ice floe distribution analysis via kernel graph cuts based feature extraction and fusion

Sakhalkar, Soumitra and Ren, Jinchang and Hwang, Phil and Murray, Paul (2015) Effective SAR image segmentation and sea-ice floe distribution analysis via kernel graph cuts based feature extraction and fusion. In: 4th International Conference on SENSOR NETWORKS, 2015-02-11 - 2015-02-13, ESEO, Angers, Loire Valley.

[img]
Preview
Text (Sakhalkar-etal-SN2015-SAR-image-segmentation-sea-ice-kernel-graph)
Sakhalkar_etal_SN2015_SAR_image_segmentation_sea_ice_kernel_graph.pdf
Accepted Author Manuscript

Download (750kB) | Preview

Abstract

The Sea Ice that grows in the open seas like the Arctic sea, forms varying shapes and size due to the fracturing as well as thickening caused by the strong gale force winds and sea waves. Over the winter season, due to the cooler temperature, these sea-ice regions combine with each other to make a stronger and larger sea ice block. In the summer however, due to the higher temperature, they separate into smaller and weaker floes. Sea-Ice monitoring has gained significant interest in recent years, largely due to the fact of the decreasing area and thickness of the older arctic sea ice (Kwok, et al., 2009) (Stroeve, et al., 2008). This decline in older sea ice has been linked largely to the growth of younger, thinner sea ice regions (Maslanik, et al., 2007) and also climate changes (Holloway & Sou, 2002), caused by greenhouse gases (Serreze, et al., 2007). The study of Polar Regions using Synthetic Aperture Radar [SAR] has been widely used for identification of sea ice floes, their size and their distribution (Burns, et al., 1987) (Rothrock & Thorndike, 1984), (Soh, et al., 2004), (Soh & Tsatsoulis, 1998). This is because SAR is not majorly affected by the harsh weather conditions or the illumination variations and it is able to cover large and primarly inaccesible areas (Xu, et al., 2014). This is particularly important for ensuring safe marine navigation as well as supporting studies of climate changes, like ours, of the Polar Regions. To date, the process of developing an automatic algorithm for effective segmentation of SAR Sea-Ice images has not been achievable. As a result, analysis of sea ice images relies on a time consuming expert analysis which is performed manually. For this reason, it is primarily important to develop techniques to automatically segment the sea-ice regions from the background and subsequently extract these sea-ice regions from the SAR image. When this is completed it will become possible to build a Floe Size Distribution (FSD) database, where FSD is a measure of the distribution of the different size of the sea ice floes. An FSD database will be constructed in our project by extracting and storing the total pixel area of these individual sea-ice regions in the SAR image and grouping them according to their size. The result will then be used to generate a graph of the size distribution of the floes on different days in a year of the Arctic region. The outcome of our study will further develop scientist’s understanding of the different trends as well as the various conditions affecting the size of the Arctic sea ice floes for that particular year. Eventually this will improve our understanding of the changes in the sea-ice extent over the year by means of comparison with the past several years’ results. The key research problem addressed by this work lies in developing a new novel image segmentation technique which is simple, fast and robust when used to segment the SAR sea ice images.