【作者】 华志励；

【导师】 顾永建； 李洪平；

【作者基本信息】 中国海洋大学 ， 海洋信息探测与处理， 2010， 博士

【摘要】 合成孔径激光雷达(SAL)是一种主动式有源成像传感器,由于其工作频率远高于微波,因此与传统的微波合成孔径雷达(SAR)相比,对于相对运动速度相同的目标可以产生更大的多普勒频移,不仅克服了普通激光雷达波束窄、搜索目标困难等缺点,而且能提供比SAR更高的方位分辨率及更短的成像时间,并具有对特定区域目标精确成像的能力。但是,在成像质量提高的同时,由于工作频率很高,SAL在成像的过程中对于大气环境中的湍流效应以及被探测目标的不规则运动非常的敏感。对此,本论文针对SAL成像过程中所涉及到的传输介质、回波信号成像处理算法以及相位误差的补偿等几个问题展开了研究,主要内容概括如下:1.提出了结构函数法生成大气湍流相位屏,并以之为基础构建大气湍流数值模型,分析传输介质中大气湍流对SAL探测信号的影响。影响SAL成像质量的原因有多个,其中非常重要的一个就是大气湍流引起的折射率起伏。针对合成孔径激光雷达信号传输过程中存在的大气湍流退化效应,本文从大气湍流的统计学特性出发,提出了一种根据不同采样点之间相位起伏的差值构建相位屏的方法——结构函数法,并以此为基础通过数值模拟的方法,建立了满足一定统计规律的大气湍流数值模型,分析了大气湍流对传输过程中光束强度特别是相位的影响机制。从实验结果可以看出,结构函数法生成的大气湍流相位屏的统计特性与理论值更为接近,一定程度上克服了现有的功率谱反演法存在的低频不足以及多项式展开法高频不足的缺陷,同时在计算效率上也有所提高。2.改进现有的线频调变标(CS)成像算法,使其适用于以连续波为信号源的合成孔径激光雷达。CS算法通过线频调技术消除了不同分辨单元之间距离徙动的空变特性,在运算量相对较小的情况下实现对距离徙动的完全校正,此特点使得该算法成为高分辨率成像算法的最优选择。同时,考虑到该算法能够适用于聚束模式的成像,因此本文选择这一算法进行分析和研究。但是,现有的CS算法是针对脉冲信号的,对于以连续波信号为信号源的SAL不完全适用。在此本文提出了两种成像处理方案,一种是通过离散采样的方法将连续波信号转换为脉冲信号,从而沿用此前基于脉冲信号的成像算法;另一种是根据连续波信号在目标斜距计算上对距离快时间处理的不同,对现有的成像算法进行改进。从实验结果来看,改进后的CS算法通过消除连续波信号回波中由快时间分量引入的频移,实现了对目标的高质量成像。3.针对SAL回波相位误差的补偿问题提出了子回波相关算法,并对现有的秩一相位估计法和盲解卷积法进行改进以获得最佳的误差补偿方案。在SAL成像过程中,大气湍流以及雷达平台的不规则运动等因素的存在不可避免地会造成SAL回波信号的相位产生误差,这对SAL高质量成像会产生严重的影响。对此,本文首先从参数模型的角度提出了子回波相关(MR)算法,利用SAL回波阵列之间的相位相关性,对存在的相位误差进行估计和补偿,一定程度上消除了相位误差,特别是低频误差的退化作用。其次,针对高频误差的补偿问题,本文对现有的两种较为成熟的非参数模型方法,即秩一相位估计(ROPE)法和迭代盲解卷积(IBD)法分别进行了改进。对于ROPE法,通过加窗处理和引入MR算法提高了图像的信噪比以及迭代初始化的精度;而对于IBD法,通过分析回波信号的多普勒信息改善了迭代解卷积盲目性的问题。实验表明,两种算法的改进措施无论对算法的精度还是效率都能起到一定的改善作用。更多还原

【Abstract】 As an active imaging sensor, the working frequency of synthetic aperture ladar (SAL) is much higher than conventional synthetic aperture radar (SAR) that works in radio domain. It results in the much improvement of both resolution and imaging time due to the achievement of higher Doppler frequency shift for targets having the same relative motions, and hence can be used to get high resolution images for certain areas. But because SAL works in a much higher frequency, it becomes much more sensitive to the working environment such as the atmospheric turbulence and the irregular targets motions. In this dissertation, three correlative topics in the process of SAL imaging are focused and discussed: the atmospheric turbulence effects on SAL backscatters, SAL imaging algorithms and phase error compensation.1. A new phase screen generating method called structure function method (SFM) is developed. Based on this SFM, a modified atmospheric turbulence model is induced to simulate the effects on SAL laser beams.Refractive index fluctuation induced by the atmospheric turbulence is one of the most important factors that limit the quality of SAL imaging. Based on the statistic characteristics of turbulence, we present here a new method of phase screen generation which is called structure function method. This method retrieves the distribution of phase fluctuation on the screen using the value of phase variation between different sampling points calculated by the wave structure function. Hence, numerical model of atmospheric turbulence can be developed to help us understand the distortion mechanism of atmospheric turbulence distortion effects on the amplitude and phase of transmitted and return laser beams. Simulation results show that a more accurate phase screen can be obtained by structure function method compared with those using traditional spectrum method and polynomial method in reasonable computation time. Low and high frequency component deficiency exist respectively in traditional spectrum and polynomial method can both be improved to certain extent. Furthermore, the computation time can be cut down due to the interval computation in SFM instead of Fourier transform in spectrum method and Zernike polynomial iteration in polynomial method.2. A suited SAL imaging algorithm modified from the traditional Chirp Scaling algorithm is presented and reported in the simulation of a SAL operated from a low altitude airborne platform.The chirp scaling (CS) algorithm is a universal method of spotlight mode SAR image processing in which the range cell migration correction operation is efficiently and accurately implemented by a range time, azimuth frequency domain multiply, thereby eliminating the traditional interpolation operation. The multiply uses the linear frequency modulation (LFM) property of the range chirp to scale the radar data in the range direction, achieving the desired range time and azimuth frequency dependent correction. Compared with pulse signal mode of traditional SAR in radio domain, continuous wave (CW) signal is a potential and practical transmitting source for SAL concentrated in this dissertation, which means original CS algorithm is not able to be completely suitable. To make the CS algorithm still available for SAL imaging, some modifications needed to be made. Here two solutions are given: one is converting the CW into pulses by discrete sampling, where the Nyquist sampling in the range direction and“neighbor return theory”in the azimuth direction are applied; the other is frequency error fore filtering ahead of azimuth fast Fourier transform (FFT), where the frequency shift needed to be eliminated is induced by fast time component in target sensor range computing. From the simulation results we can see, improvement of image resolution shows the effectiveness of such modifications.3. Phase error compensation algorithms that show the improvement of both accuracy and efficiency are presented to help realize a better solution of SAL imaging.The distortion of atmospheric turbulence and unforeseen vibration of sensor platform would induce phase error in laser beam, and hence seriously degrade the quality of SAL imaging. Based on the parametric model of phase error, or the point spread function (PSF) of turbulence, map return (MR) algorithm is reported here to estimate the phase error existed in each return by the correlation information contained in phase history data (PHD). On the other hand, as two previously devised methods, rank one phase error estimation (ROPE) algorithm and iterative blind deconvolution (IBD) are reexamined here, to which some modifications are given to be more effective. For ROPE algorithm, windowing and MR pre estimation can both improve the image signal to noise ratio and initialization accuracy. For IBD algorithm, Doppler information of backscatter signals can be used to find the suited iteration times. Simulation results show that modified SAL image processing algorithm can give us a satisfied solution in both computing accuracy and efficiency.更多还原