Research on Weak Target Recognition of Oil Film in Ice Based on Dual-Polarized SAR

Given the similar low backscatter coefficients of oil spills and sea ice in SAR images, it is crucial to determine which features can effectively distinguish between sea ice and oil film. This is essential for accurate oil spill detection in icy waters. Consequently, this paper explores two types of features: polarization target decomposition and grayscale co-occurrence matrix. Using the J-M distance metric to assess the separability of ice and oil in feature images, selecting various feature combinations, and exploring combinations suitable for distinguishing oil film from sea ice, the SVM algorithm is employed to classify the ice and oil samples. This study demonstrates that the polarization target decomposition feature offers significant advantages in distinguishing between sea ice and oil film. Additionally, it serves as a valuable reference for monitoring oil spills in icy sea areas.