首页|基于精细极化分解的PolSAR车辆目标检测方法

基于精细极化分解的PolSAR车辆目标检测方法

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
极化合成孔径雷达(polarimetric synthetic aperture radar，PolSAR)能够获取目标丰富的极化信息，在人造目标的散射机理解译和检测、分类、识别等领域发挥着重要的作用。PolSAR车辆目标检测对环境侦察监视、道路交通监管等应用来说尤为关键。不同于海上的舰船目标，车辆目标所处的地面背景包括草地、树木、建筑、道路等，背景杂波较为复杂，人造杂波干扰较强，这给PolSAR车辆目标的检测带来了 一定的挑战。车辆目标散射机理的准确解译是后续极化特征提取和目标检测方法设计的基础。为了准确描述车辆目标的散射行为，本文构建了旋转二面角散射模型，并由此提出了一种精细五分量极化分解方法。在此基础上，通过分析车辆目标与其地面背景的散射特性差异，本文提出了一种散射功率复合特征检测器，用于车辆目标的检测。在两景实测无人机载SAR数据上的实验结果证明，本文方法能够显著提高车辆目标处的目标杂波比(target-to-clutter ratio，TCR)，获得了优异的检测性能。此外，本文方法能够保留车辆目标完整的轮廓和细节信息，有助于后续的分类、识别等应用。
PolSAR vehicle detection based on refined polarimetric decomposition
Polarimetric synthetic aperture radar(PolSAR)can acquire full target polarization information,playing an important role in man-made target scattering mechanism interpretation and in target detection,classification,and recognition,etc.Vehicle detection of PolSAR image is especially critical for applications such as environmental surveillance and traffic management.Different from ships on the sea,the terrain backgrounds of vehicles are generally more complex,consisting of grass,trees,buildings,roads,etc.These result in complicated background clutters and intense artificial clutter interferences,which pose major challenges for PolSAR vehicle detection.It is widely known that accurate interpretation of vehicle scattering is the momentous foundation for polarimetric feature extraction and target detection.To accurately characterize the vehicle scattering behaviors,we construct a rotated dihedral scattering model(RDSM),and then propose a refined five-component decomposition method.Afterward,through analyzing the scattering differences between vehicles and their terrain backgrounds,we present a scattering power composite detector for PolSAR vehicle detection.The experimental results on two measured UAV-borne SAR datasets demonstrate the proposed method can significantly enhance the target-to-clutter ratios(TCR)of vehicles and achieve superior detection performance.In addition,it can remain complete contours and details of vehicles,contributing to subsequent target classification,recognition and so on.

polarimetric synthetic aperture radar(PolSAR)vehicle detectionrotated dihedral scatteringrefined polarimetric decompositionscattering power composite feature

全斯农、段颜翠、王威、邢世其、徐振海、肖顺平

展开 >

国防科技大学电子科学学院电子信息系统复杂电磁环境效应国家重点实验室,长沙 410073

国防科技大学电子科学学院自动目标识别国防科技重点实验室,长沙 410073

极化合成孔径雷达 车辆目标检测 旋转二面角散射 精细极化分解 散射功率复合特征

国家自然科学基金青年科学基金博士后创新人才支持计划

62001487BX20220400

2024

10.1360/SSI-2023-0337

中国科学F辑
中国科学院,国家自然科学基金委员会

中国科学F辑

CSTPCD北大核心
影响因子：1.438
ISSN：1674-5973
年,卷(期)：2024.54(6)