目标三维电磁散射参数化模型反演方法研究
发布时间：2018-11-06 14:52
【摘要】：目标电磁散射建模是基于模型的雷达目标识别系统的核心内容之一。目标三维电磁散射参数化模型,特别是基于典型散射结构的参数化模型,用一组简洁的参数描述目标,能够为识别系统提供物理意义明确的多层次目标特征。从电磁散射测量数据中反演目标参数化模型是一个具有挑战性的任务。论文围绕从多角度合成孔径雷达数据中建立目标三维散射特性参数化模型的问题,研究了反演框架以及其中的多个关键问题。在阐述清楚目标三维电磁散射参数化模型反演的内涵、研究内容和面临挑战的基础上,论文第二章提出了一种基于典型散射结构的目标参数化模型反演框架,反演参数的物理意义更加清晰。该框架由模型初始化和参数优化两部分组成,利用多角度合成孔径雷达数据反演目标参数化模型,具有较强的灵活性,便于综合运用多种技术途径完成建模任务。针对模型初始化问题,论文重点研究了目标二维/三维散射中心特征提取方法。基于稀疏表示与压缩感知理论,论文第三章提出了一种二维散射中心提取方法,论文第四章提出了雷达目标三维成像方法和三维散射中心提取方法。所提方法利用目标图像的先验信息和模型时域响应的特点降低稀疏重构的维度和数据量,在模型维度较高的情况下仍可保证较高的效率。论文第五章提出了基于位置聚类分析、散射中心参数匹配、压缩感知等三种利用多个二维散射中心重构三维散射中心的方法,这些方法降低了对多角度数据的要求,适应处理宽基线多角度合成孔径雷达数据,重构结果与目标结构对应性较好。针对目标典型散射结构参数化模型的参数优化问题,论文第六章分别从图像域约束准则和多角度图像分割两个角度提出了优化方法,扩大收敛范围,提高了参数优化稳健性。基于上述反演框架和关键方法,论文第七章提出了典型散射结构和点散射模型相结合的复杂目标参数化建模方法,实现了目标三维电磁散射参数化模型反演原型系统,利用目标的电磁计算数据反演了目标全方位角-大俯仰角的三维电磁散射参数化模型的反演,并分析了模型的精度。实验结果验证了所提框架和方法的可行性及有效性。
[Abstract]:Target electromagnetic scattering modeling is one of the core contents of radar target recognition system based on model. The three-dimensional electromagnetic scattering parameterized model of target, especially the parameterized model based on typical scattering structure, describes the target with a set of simple parameters, which can provide the multi-level target characteristics with clear physical meaning for the recognition system. It is a challenging task to retrieve the target parameterized model from the electromagnetic scattering measurement data. In this paper, a parameterized model of 3-D scattering characteristics from multi-angle synthetic aperture radar (SAR) data is established, and the inversion framework and several key problems are studied. On the basis of explaining the connotation, research content and challenge of 3D electromagnetic scattering parameterized model inversion of target, in the second chapter, a target parameterized model inversion framework based on typical scattering structure is proposed. The physical meaning of the inversion parameters is clearer. The framework consists of two parts: model initialization and parameter optimization. The multi-angle synthetic aperture radar data is used to retrieve the target parameterized model. It has strong flexibility and is convenient to complete the modeling task by comprehensive use of various technical approaches. Aiming at the problem of model initialization, this paper focuses on the feature extraction method of two-dimensional / three-dimensional scattering centers. Based on sparse representation and compressed sensing theory, a two-dimensional scattering center extraction method is proposed in the third chapter. In the fourth chapter, the radar target three-dimensional imaging method and three-dimensional scattering center extraction method are proposed. The proposed method can reduce the sparse reconstruction dimension and the amount of data by using the prior information of the target image and the time-domain response of the model. The proposed method can still ensure a higher efficiency when the model dimension is high. In the fifth chapter, three methods based on location clustering analysis, scattering center parameter matching and compression perception are proposed to reconstruct 3D scattering center using multiple two-dimensional scattering centers. These methods reduce the requirement of multi-angle data. It adapts to the wide baseline multi-angle synthetic aperture radar (SAR) data, and the reconstruction results correspond well with the target structure. Aiming at the parametric optimization problem of typical scattering structure, in chapter 6, the optimization method is proposed from two angles of image domain constraint criterion and multi-angle image segmentation, which expands the convergence range and improves the robustness of parameter optimization. Based on the above inversion framework and key methods, in chapter 7, a complex target parameterized modeling method combined with typical scattering structure and point scattering model is proposed, and a prototype system of 3D electromagnetic scattering parameterized model inversion is implemented. The inversion of the three-dimensional electromagnetic scattering parameterized model with omni-directional angle and large pitch angle is obtained by using the electromagnetic calculation data of the target, and the accuracy of the model is analyzed. The experimental results verify the feasibility and effectiveness of the proposed framework and method.
【学位授予单位】：国防科学技术大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TN957.52
，
本文编号：2314605
[Abstract]:Target electromagnetic scattering modeling is one of the core contents of radar target recognition system based on model. The three-dimensional electromagnetic scattering parameterized model of target, especially the parameterized model based on typical scattering structure, describes the target with a set of simple parameters, which can provide the multi-level target characteristics with clear physical meaning for the recognition system. It is a challenging task to retrieve the target parameterized model from the electromagnetic scattering measurement data. In this paper, a parameterized model of 3-D scattering characteristics from multi-angle synthetic aperture radar (SAR) data is established, and the inversion framework and several key problems are studied. On the basis of explaining the connotation, research content and challenge of 3D electromagnetic scattering parameterized model inversion of target, in the second chapter, a target parameterized model inversion framework based on typical scattering structure is proposed. The physical meaning of the inversion parameters is clearer. The framework consists of two parts: model initialization and parameter optimization. The multi-angle synthetic aperture radar data is used to retrieve the target parameterized model. It has strong flexibility and is convenient to complete the modeling task by comprehensive use of various technical approaches. Aiming at the problem of model initialization, this paper focuses on the feature extraction method of two-dimensional / three-dimensional scattering centers. Based on sparse representation and compressed sensing theory, a two-dimensional scattering center extraction method is proposed in the third chapter. In the fourth chapter, the radar target three-dimensional imaging method and three-dimensional scattering center extraction method are proposed. The proposed method can reduce the sparse reconstruction dimension and the amount of data by using the prior information of the target image and the time-domain response of the model. The proposed method can still ensure a higher efficiency when the model dimension is high. In the fifth chapter, three methods based on location clustering analysis, scattering center parameter matching and compression perception are proposed to reconstruct 3D scattering center using multiple two-dimensional scattering centers. These methods reduce the requirement of multi-angle data. It adapts to the wide baseline multi-angle synthetic aperture radar (SAR) data, and the reconstruction results correspond well with the target structure. Aiming at the parametric optimization problem of typical scattering structure, in chapter 6, the optimization method is proposed from two angles of image domain constraint criterion and multi-angle image segmentation, which expands the convergence range and improves the robustness of parameter optimization. Based on the above inversion framework and key methods, in chapter 7, a complex target parameterized modeling method combined with typical scattering structure and point scattering model is proposed, and a prototype system of 3D electromagnetic scattering parameterized model inversion is implemented. The inversion of the three-dimensional electromagnetic scattering parameterized model with omni-directional angle and large pitch angle is obtained by using the electromagnetic calculation data of the target, and the accuracy of the model is analyzed. The experimental results verify the feasibility and effectiveness of the proposed framework and method.
【学位授予单位】：国防科学技术大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TN957.52
，
本文编号：2314605
本文链接：https://www.wllwen.com/shoufeilunwen/xxkjbs/2314605.html
