基于极化SAR分解模型的油膜检测

Spill detection based on polarimetric SAR decomposition models

为了更好地利用极化合成孔径雷达(SAR)检测海面溢油,该文提出了基于极化分解模型的特征——Bragg散射能量占比。在油膜区域,Bragg散射能量占比会比较小;而在海面和常见油膜类似物如油醇(OLA)区域,该特征的值会比较大。因此,该特征能够有效地从海杂波里检测出油膜区域,并且能够有效地排除OLA虚警。并且该特征可以被推广应用于HH\VV双极化数据和紧缩极化(CP)模式,从而可以进行大面积的溢油监测。利用C波段的极化SAR数据的实验表明:在各种极化模式下该特征能够检测出油膜,并有效鉴别油膜和油膜类似物。

Spill detection using polarimetric SAR is proposed by using Bragg scattering energy proportion based on the polarimetric decomposition model.The Bragg scattering energy proportion is small in a spill region but larger in sea clutter or an Oleyl alcohol（OLA）region.Thus,this feature can detect oil spills from sea clutter while excluding false OLA alarms.The feature can be easily extended to HH／VV dual-polarimetric and compact polarimetric configurations which give wider radar coverage.Experiments with multi-look complex C band SAR data show that for all polarimetric modes,the feature gives state-of-art spill detection and false-alarm exclusion of spills.