Automatic classification of RADARSAT SAR images of the Northern Sea Route

Two automatic sea ice classification algorithms based on linear discriminant analysis (LDA) and feedforward backpropagation multilayer perceptron (MLP) were compared for classification of RADARSAT ScanSAR images of the Western part of the Northern Sea Route. Texture and local statistical parameters in addition to radar image brightness are used as input features to the algorithms. Simultaneous in-situ observations were carried out during an expedition to the Barents and Kara seas afforded to assess the performance of the algorithms.