Ren, P.; Yu, Z.-Q.; Dong, G.S.; Wang, G.-X., and, Wei, K., 2019. Sea ice classification with first-order logic refined sliding bagging. In: Jung, H.-S.; Lee, S.; Ryu, J.-H., and Cui, T. (eds.), Advances in Remote Sensing and Geoscience Information Systems of Coastal Environments. Journal of Coastal Research, Special Issue No. 90, pp. 129-134. Coconut Creek (Florida), ISSN 0749-0208.

This paper proposes an automatic framework for classifying sea ice types using remote sensing imageries. Firstly, polarization features are extracted to form polarization characteristic vectors for representing individual pixels in a remote sensing image. Secondly, a multiple classifier ensemble strategy, i.e., sliding bagging, is exploited for classifying each polarization characteristic vector into a sea ice type. The sliding bagging strategy not only avoids the limitation of one single classifier for characterizing the sea ice variability over a large-scale region, but also alleviates the data imbalance over different sea ice types. Finally, the spatial structural relationships of sea ice types are encoded with first-order logic (FOL) to refine the sea ice classification resulted from the sliding bagging scheme. Experimental evaluations on RADARSAT-2 data validate the effectiveness of the proposed framework.