Ultra-High Resolution KOMPSAT-5 SAR Sea-Ice Random Radiometric Anomaly Formulation Based On Hybrid Domain Texture Decomposition

摘要

Image texture in synthetic aperture radar (SAR) is a mixture of inherent texture of the surface backscattering and non-coherent backscattering interactions in the form of random radiometric anomalies that ends up in the loss of information and data inconsistency. According to the multilayer structure of polar region, which is a non-deterministic and constantly changing medium on the basis of time, these time- dependent anomalies can lead to misinterpretation. As a novel approach, a time-frequency Gamma probability distribution function (PDF) formulation of such multiplicative anomaly has been proposed for high-resolution SAR sea-ice dataset information analysis and the texture quality improvement. The proposed hybrid-domain approach is tested on a real KOMPSAT-5 SAR satellite sea-ice image and the results verify the formulation, modeling, and their suppression scenario.