【作者】 高勋章；

【导师】 庄钊文；

【作者基本信息】 国防科学技术大学 ， 信息与通信工程， 2004， 博士

【摘要】 本文研究了基于高阶统计量的雷达目标高分辨成像技术。 首先回顾了雷达成像技术的发展历史和研究现状,简要概述了基于高阶统计量的谐波恢复问题,并指出了目前存在的主要问题。 第二章研究了基于乘积性高阶模糊函数的空间目标一维距离像运动补偿方法。推导了平动和转动状态下目标雷达回波的多分量多项式相位信号模型,分析了高速平动和转动对空间目标一维距离像的影响。给出了两类不同稳定状态的空间目标运动估计和补偿方法,并详细分析了影响模型参数估计精度的各种因素。 第三章针对现有成像方法对高斯色噪声敏感的问题,研究了高斯色噪声背景下基于四阶混合累积量的雷达目标一维超分辨成像。首先对谐波信号的混合累积量概念进行了推广,给出了衰减谐波的混合累积量的定义并证明了其渐近盲高斯特性。利用超分辨谱估计技术,提出了基于四阶混合累积量对角切片的FOMCMP和FOMCESPRIT方法,分析了这两种方法的参数估计性能。最后给出实测数据的成像实验结果。 第四章研究了高斯色噪声背景下基于二维四阶混合累积量的雷达目标二维超分辨成像。定义了衰减谐波的二维混合累积量,提出了二维衰减谐波信号参数估计的2d-FOMCESPRIT方法,分析了这种方法的计算量和参数估计性能。最后给出仿真数据和实测数据的成像实验结果。结果表明二维四阶混合累积量不仅能够降低雷达成像的高斯色噪声敏感性,而且可以大大减小成像算法的计算量。 第五章针对真实信号中包含任意平稳加性和乘性噪声的情况,研究了基于高阶循环统计量的雷达目标成像。提出了乘积型和累加型循环统计量的概念,并从理论上证明了乘积型和累加型循环矩(累积量)估计子的渐近特性。针对多分量随机幅度和多分量时变幅度模型,重点研究了基于乘积型循环均值的模型参数估计方法。实验表明该方法的抗噪能力和可辨识性优于传统方法。 最后对全文的工作进行了总结,指出了需要进一步深入研究和解决的问题。更多还原

【Abstract】 The research presented investigates the use of higher-order statistics to the imaging techniques of high resolution radar targets.After briefly reviewing the developments of radar imaging and harmonic retrieval based on higher-order statistics, the first chapter discusses the main problems in these areas and introduces the main contents of this dissertation.Motion compensation of space target is studied in chapter 2. Firstly the influence on HRRP (high resolution range profile) of space target because of its translation and rotation is analyzed after the mc-PPS model of radar echoes is derived. Based on product high-order ambiguity function (PHAF), two methods to compensate the motion are proposed.Super resolution imaging methods for one-dimensional imaging is studied using fourth-order mixed c umulants (FOMC) in chapter3.At first, we extended the FOMC of sinusoids to damped exponential model (DE) and prove its asymptotic blind Gaussian noise character, then two efficient methods named FOMCMP and FOMCESPRIT for estimating the parameter of DE model in the presence of colored Gaussian noise are proposed. Monte Carlo simulations are demonstrated lastly.In the next chapter 4, the definitions and methods are extended to the two-dimensional (2d) ISAR imaging. Similar results of blind Gaussian noise character of 2d-FOMC are derived. The computation efficiency of the proposed method named 2d-FOMCESPRIT is greatly improved by holding the feature matrix’s dimension in the condition of a long data record. Simulation results show that 2d-FOMC can suppress the colored Gaussian noise more effectively and has higher efficiency than former super resolution methods.In chapter 5, the cyclic statistics based radar imaging is discussed in the presence of additive and multiplicative stationary noise with any statistic distribution. The concepts of product and cumulate cyclic statistics are introduced and their asymptotic character is proved. A new approach based on product cyclic average for multi-component harmonic retrieval with random or time-varying amplitudes is presented. Finally, tests based on simulation data and real data validate the new method.At last, Summary of this dissertation is made and the problems which need further research are pointed out in chapter 6.更多还原