基于对数空间理论的极化SAR图像舰船目标检测新方法研究

The polarimetric synthetic aperture radar (Pol-SAR) is an important information source in civil earth observation and military intelligence collection for its multi-dimensional ground remote sensing capability. It is a hot and difficult problem in the imagery interpretation of Pol-SAR images how to detect interesting targets in the complicated clutter. Our project deals with the theoretical framework of the polarization information processing methods by using the logarithmic space and the creative application to target detection in Pol-SAR images. The clutter in complicated background is regarded as polarimetric L distribution, polarimetric G distribution or polarimetric heavy-tailed Raleigh distribution. To show the differences of the characteristics between targets and clutters, the matrix logging method is adopted, which establishes the logarithm space of polarization information processing. It is explained why the matrix logarithm has the advantage here. Therefore a novel estimation method based on the logarithmic space of polarization information processing is proposed, so is an effective distinguishing method of different random distributed clutters. Furthermore, a new noise filtering method of the multivariate product model is obtained through the additive noise transformation. Thus, the systematic theory framework of the polarization information processing can be constructed according to the matrix logarithm theory. Then a novel target decomposition method of polarization called synthetic target decomposition is obtained based on the incoherent and coherent target decomposition methods of polarization. The spatial instantaneous polarization characteristics of ships and ocean are analylized,so is the optimization of sub-band splitting.The theory of logarithmic space is used to the creative target detection by means of clutter models,the polarimetric White and Matched filtering, synthetic target decomposition and purity of polarization and spatial instantaneous sub-aperture decompostion theory based on the derivations of the probability density function (PDF) of the characteristics of targets and clutters. Finally, novel responding methods for target detection are presented, which provide a key technical support for interesting target detection in the areas of civil remote sensing and military observation.

极化合成孔径雷达是民用海洋观测和军事情报获取的重要信息源。如何从复杂海洋背景中检测出舰船目标是极化SAR图像解译研究领域的热点和难点问题。 本项目针对极化L分布、极化G分布和重尾瑞利分布等多变量乘积模型的复杂海洋背景，以凸显目标特征和优化数据处理为目的，采用极化协方差矩阵对数变换的方法，构建极化对数空间理论框架；然后探究基于极化对数空间的随机模型有效辨识方法、快速参数估计方法和加性滤波模型，揭示极化对数空间潜在优势的本质；最后提出有机衔接非相干极化分解和相干极化分解理论的融合极化分解理论，推导舰船目标空域瞬态极化特性和最佳子带划分方法，建立基于极化统计模型、极化白化匹配滤波、散射随机性、融合极化分解理论和空域瞬态极化子孔径分解理论的极化SAR舰船目标检测新原理新方法。本研究既可为较为分散的SAR对数处理方法建立系统理论，也能为民用遥感与军事观测领域的极化SAR舰船目标检测提供关键技术支撑。

极化合成孔径雷达是民用海洋观测和军事情报获取的重要信息源。如何从复杂海洋背景中检测出舰船目标是极化SAR图像解译研究领域的热点和难点问题。.本项目首先针对极化L分布、极化G分布和重尾瑞利分布等多变量乘积模型的复杂海洋背景，以凸显目标特征和优化数据处理为目的，采用极化协方差矩阵对数变换的方法，构建雷达极化对数空间；其次探究基于极化对数空间的随机分布有效辨识方法和快速参数估计方法，揭示极化对数空间潜在优势的本质；最后提出基于极化统计模型、极化白化匹配滤波、散射随机性和空域瞬态极化子孔径分解理论的极化SAR舰船目标检测新原理新方法。本研究既可为目前较为分散的SAR对数处理方法建立系统理论，也能为民用遥感与军事观测领域的极化SAR舰船目标检测提供关键技术支撑。