基于简缩极化SAR的溢油检测与分类方法

Oil Spill Detection and Classification Method Based on Compact Polarization SAR

针对简缩极化SAR在海上溢油的检测与分类应用开展研究,利用欧式距离全面分析了简缩极化SAR的36种极化特征在溢油检测与油膜分类中的性能,发现简缩极化特征中的奇次散射系数的溢油检测性能最好,简缩极化熵的疑似溢油鉴别性能最好。在此基础上,提出了结合二叉树原理的简缩极化SAR溢油检测与油膜分类算法,并分析了RADARSAT-2和SIR-C全极化溢油数据模拟的简缩极化数据。结果表明,此方法对溢油的检测精度可达95.67%,对于疑似溢油的识别精度可达95.71%,证明了简缩极化SAR在溢油检测与分类中具有较好的应用前景。

The application of compact polarization SAR in the detection and classification of oil spill was analyzed in this study.First,we perfomed a comprehensive analysis of the performance of 36 features of compact polarization SAR in oil spill detection and classification by using Euclidean distance,and found that the odd-order scattering coefficient had best performance on oil spill detection,and the compact polarization entropy achieved better performance on oil spill lookalikes identification.Then,we proposed a compact polarization SAR oil spill detection and classification algorithm based on the binary tree idea,and the fully-polarimetric data of RADARSAT-2 and SIR-C were used to reconstruct the reduced polarized data and experiments.Results showed that the detection accuracy of oil spill detection is 95.67%,and the identification accuracy of oil spill lookalikes can reach 95.71%,which is 2%higher than the classic Wishart supervised classification method,indicating that the compact polarization SAR has a better application prospect in oil spill detection and classification.