基于极化合成孔径雷达数据的桥梁检测

为了对遥感合成孔径雷达(SAR)图像中桥梁目标进行自动检测,在极化散射矩阵参数的基础上,提出了一种寻找河流地区对比参数的方法:将总功率、极化熵、相似性参数等参数进行加权组合,利用特征值分析方法求取最优加权系数,从而得到对河流特征敏感的参数组合表达式。进而利用该表达式对河流地区进行了分类实验,对分类结果应用图形形态学方法处理后形成水域模板。在此基础上将水域模板对敏感参数图像覆盖后,对河上桥梁目标进行了自动检测。采用PI-SAR型的雷达的全极化数据进行实验,获得了良好的检测效果,和只使用功率数据进行桥梁检测相比,原来检测不出来的桥梁目标都明显显现出来。实验结果表明,综合利用全极化SAR信息在遥感应用中可以更好的检测和识别目标。

Highway bridge detection based on polarimetric SAR data

A highway bridge detection method was developed for synthetic aperture radar (SAR) remote sensing data based on backscattering polarimetric parameters such as span, entropy, and similarity. The optimal coefficients for a linear combination equation describing the bridge detection were obtained by solving the eigenvalue problem for the system. The river area is classified using a morphological method to build a template of the river geometry to automatically detect the bridges across the river. The system worked effectively with PI-SAR polarimetric data to efficiently recognize bridges.