Sea ice change detection from sar images based on canonical correlation analysis and contractive autoencoders

Abstract

In this paper, we proposed a novel sea ice change detection method for Synthetic Aperture Radar (SAR) images based on Canonical Correlation Analysis (CCA) and Contractive Autoencoders (SCAEs). To alleviate the effect of multiplicative speckle noise, structured matrix decomposition is utilized for difference image enhancement, and therefore, better difference image with less noisy spots can be obtained. In order to get good data representations in changed and unchanged pixels classification, CCA and SCAEs are combined to exploit more effective changed features. Experiments on two real sea ice datasets demonstrate the robustness and efficiency of the proposed method in comparison with three other state-of-the-art methods.