| 摘要: |
| 为完成极化合成孔径雷达(PolSAR)图像中点散射像素目标的检测，从几何扰动滤波检测点散射像素的基本原理出发，分析了经典方法存在的相干度参数阈值无法自适应获取和不同散射机制共享同一阈值两个问题，提出利用Cloude-Pottier分解结果中熵参数的分布情况计算得到阈值，利用平均阿尔法角参数完成不同散射机制的初分类，对初分类的结果排序得到的某种散射机制对应的相干度参数的比例因子，根据比例因子计算得到该类散射机制的阈值，从而完成点散射目标的检测。对机载合成孔径雷达（AIRSAR）数据集中的San Francisco Bay图像进行了实验，结果表明，改进方法在检测性能上优于经典方法。 |
| 关键词: 极化合成孔径雷达 点散射目标检测 几何扰动滤波 熵参数 平均阿尔法角 |
| DOI： |
|
| 基金项目:国家自然科学基金资助项目（61202269）；贵州省科学技术基金项目(黔科合LH字［2014］7206) |
|
| An improved point target detection algorithm based on geometric perturbation filter in polarimetric SAR images |
| SUN Sheng,XU Zhijia,LIU Renfeng |
| () |
| Abstract: |
| For achieving the point target detection in polarimetric synthetic aperture radar(PolSAR) images, two problems of classic method,including the adaptive acquiring of coherence threshold and exclusive threshold acquiring,are analyzed according to the principle of geometric perturbation filter. The distribution of entropy parameter,produced by the Cloude-Pottier decomposition,is employed to calculate the threshold. The averaged alpha angle is employed to make a preliminary classification. The ratio of coherence values can be deduced from the sorting results of a specific scattering mechanism. The threshold can be obtained from the ratio so that the detection of singe target is ultimately accomplished. The experiment performed on the sample image San Francisco Bay in AIRSAR datasets shows that the improved method overmatches the classic method in performance. |
| Key words: polarimetric SAR point target detection geometric perturbation filter entropy parameter averaged alpha angle |
《电讯技术》编辑部 