微波遥感基于模型目标分解的研究

Polarimetric synthetic aperture radar(POLSAR) is an important branch of synthetic aperture radar(SAR). As model-based target decomposition can interpret the scattering mechanism between a target and the incident polarimetric wave, it is the foundation of polarimetric classification, target recognition, parameters inversion and other applications. Nevertheless, the existing algorithms have some errors in understanding target scattering mechanisms. There appear unreasonable phenomena in interpreting the scattering mechanisms of buildings, such as volume scattering power overestimation, and the appearance of negative power pixels, which affect the applications of POLSAR. Respect to this scientific problem, firstly, starting from the aspect of the target is a polarimetric transformer to polarimetric wave, the research extracts two scattering features, i.e., target ellipticity angle and target phase angle by utilizing scattering characteristics and polarimetric basis transformation, diggs their physical properties and develops their target compensations. Meanwhile, respect to the influence of multilook process, the research proposals a distributed surface scattering model and a distributed dihedral scattering model, which are the functions of normalized circular-pol correlation coefficients and equivalent number of looks(ENL).Then, the research presents an improved target decomposition which contains target compensations and considers the impact of multilook process. At last, the research develops an improved scattering feature-preserving classification method to validate the scattering features and the improved target decomposition. The research can enrich the existing polarimetric scattering theory. In other words, target ellipticity angle and target phase angle can describes the process between a target and its incident microwave. Further, this research also can offer reliable data and critical technique for wide applications, for example, city planning, land cover classification, and topographic mapping.

极化SAR是SAR的重要分枝。基于模型目标分解能解译目标和电磁波的作用机制，是极化SAR分类、识别、参数反演等运用的基础。现有算法对目标散射机制认识存在误差，建筑出现体散射功率估计过高、负功率像素的不合理现象，影响极化SAR的运用。针对这一科学问题，研究首先从目标对电磁波变极化作用的角度出发，利用散射特性和极化基转换从相干矩阵提取目标椭圆率角和目标相位角，挖掘对应物理性质和发展对应目标补偿。同时，针对多视处理影响，研究提出包含归一化圆极化相关系数、等效视数的分布面散射模型和分布二面角散射模型。之后，提出一种包含目标补偿和考虑多视处理影响的改进目标分解算法。最后，提出一种散射特性保持的分类方法对散射特征和改进目标分解进行验证。开展本研究能充实目标极化散射理论体系——目标椭圆率角和目标相位角能阐述目标与电磁波的作用机理，而且也能为城市规划、土地覆盖分类、地形测绘等运用提供科学数据和技术支撑。

极化SAR是SAR的重要分枝。基于模型目标分解能解译目标和电磁波的作用机制，是极化SAR分类、识别、参数反演等运用的基础。现有算法对目标散射机制认识存在误差，建筑出现体散射功率估计过高、负功率像素的不合理现象，影响极化SAR的运用。针对这一科学问题，研究首先从目标对电磁波变极化作用的角度出发，利用散射特性和极化基转换从相干矩阵提取目标椭圆率角和目标相位角，挖掘对应物理性质和发展对应目标补偿。同时，针对多视处理影响，研究提出包含归一化圆极化相关系数、等效视数的分布面散射模型和分布二面角散射模型。之后，提出一种包含目标补偿和考虑多视处理影响的改进目标分解算法。最后，提出一种散射特性保持的分类方法对散射特征和改进目标分解进行验证。开展本研究充实了目标极化散射理论体系——目标椭圆率角和目标相位角能阐述目标与电磁波的作用机理，而且也为城市规划、土地覆盖分类、地形测绘等运用提供科学数据和技术支撑。课题已经发表了6篇文章，其中2篇SCI，申请了2个专利，培养了1名博士和4名研究生，研究成果达到了结题要求。