Chapter 6
Mapping Sea Ice from Satellite SAR Imagery
Linlin Xu and Jonathan Li
Abstract Sea ice information is crucial for ensuring safe marine navigation and
supporting climate change studies in the polar regions. Spaceborne synthetic
aperture radar (SAR), due to its ability to cover large inaccessible areas without
the dependence on weather condition or sun-light illumination, provides a powerful
tool for sea ice mapping. This chapter provides a comprehensive overview of SAR
image analysis approaches to sea ice mapping with a focus on sea ice segmentation.
Sea ice segmentation is an essential step in computer-aided sea ice mapping
systems. Automated segmentation of SAR sea ice imagery is a difficult task due
to the complex sea ice physics and the ever-changing ocean environment, as well as
the numerous sensor parameters. In light of the difficulties, an efficient segmentation method has to utilize the spatial and textural information for modeling the label
correlation and increasing the discriminative capability. This Chapter presents a
Bayesian method for segmentation of SAR sea ice imagery, where a novel kernel
principal component analysis (KPCA) model is used for accounting for the textual
information, and a Markov random filed (MRF) is used for addressing the label
correlation effect. The proposed method is optimized by the graph-cut approach.
The results demonstrate that the proposed method is capable of effectively delineating different sea ice types. Moreover, it requires less computational time than the
other advanced approaches.
Keywords Sea ice segmentation • Synthetic Aperture Radar (SAR) • Bayesian
estimation • Maximum a posterior • Markov random field (MRF) • Kernel principal
component analysis
L. Xu • J. Li (*)
Department of Geography and Environmental Management, University of Waterloo,
Waterloo, ON N2L 3G1, Canada
e-mail: l44xu@uwaterloo.ca; junli@uwaterloo.ca
© Springer Science+Business Media Dordrecht 2015
J. Li, X. Yang (eds.), Monitoring and Modeling of Global Changes:
A Geomatics Perspective, Springer Remote Sensing/Photogrammetry,
DOI 10.1007/978-94-017-9813-6_6
113
Mapping Sea Ice from Satellite SAR Imagery
Linlin Xu and Jonathan Li
Abstract Sea ice information is crucial for ensuring safe marine navigation and
supporting climate change studies in the polar regions. Spaceborne synthetic
aperture radar (SAR), due to its ability to cover large inaccessible areas without
the dependence on weather condition or sun-light illumination, provides a powerful
tool for sea ice mapping. This chapter provides a comprehensive overview of SAR
image analysis approaches to sea ice mapping with a focus on sea ice segmentation.
Sea ice segmentation is an essential step in computer-aided sea ice mapping
systems. Automated segmentation of SAR sea ice imagery is a difficult task due
to the complex sea ice physics and the ever-changing ocean environment, as well as
the numerous sensor parameters. In light of the difficulties, an efficient segmentation method has to utilize the spatial and textural information for modeling the label
correlation and increasing the discriminative capability. This Chapter presents a
Bayesian method for segmentation of SAR sea ice imagery, where a novel kernel
principal component analysis (KPCA) model is used for accounting for the textual
information, and a Markov random filed (MRF) is used for addressing the label
correlation effect. The proposed method is optimized by the graph-cut approach.
The results demonstrate that the proposed method is capable of effectively delineating different sea ice types. Moreover, it requires less computational time than the
other advanced approaches.
Keywords Sea ice segmentation • Synthetic Aperture Radar (SAR) • Bayesian
estimation • Maximum a posterior • Markov random field (MRF) • Kernel principal
component analysis
L. Xu • J. Li (*)
Department of Geography and Environmental Management, University of Waterloo,
Waterloo, ON N2L 3G1, Canada
e-mail: l44xu@uwaterloo.ca; junli@uwaterloo.ca
© Springer Science+Business Media Dordrecht 2015
J. Li, X. Yang (eds.), Monitoring and Modeling of Global Changes:
A Geomatics Perspective, Springer Remote Sensing/Photogrammetry,
DOI 10.1007/978-94-017-9813-6_6
113