【作者】 索志勇；

【导师】 保铮；

【作者基本信息】 西安电子科技大学 ， 信号与信息处理， 2008， 博士

【摘要】 合成孔径雷达干涉测量是雷达遥感测量的研究热点,根据其工作方式可分为垂直航迹干涉合成孔径雷达(Cross-Track Interferometric Synthetic Aperture Radar,XTI)和沿航迹干涉合成孔径雷达(Along-Track Interferometric Synthetic Aperture Radar, ATI)两类,它利用两幅或两幅以上具有相干性的SAR复图像进行联合处理,相比于传统的单幅合成孔径雷达(SAR)图像能够获得更多的信息,利用垂直航向基线可以获得地面高精度的数字高程图,利用沿航向的基线可以进行地面运动目标检测以及洋流监测等,因而在军事、国民经济建设和科学研究中,有着极其广泛的应用领域。SAR图像的分辨率越来越高,体散射、多径传播等会影响干涉SAR的相位质量,而且雷达平台的系统测量设备的位置精度通常不能满足要求,因而必须通过各种手段来提高参数估计精度和数据处理速度,同时尽可能使一个系统能够具有多种功能,做到一个设备多种用途。本文结合实际数据处理中遇到的问题,对垂直航迹和沿航迹干涉的数据处理方法进行了一些探索。本文的主要工作概括如下:1.提出了一种利用全局最小二乘拟合的局部梯度估计方法,用于获取相位滤波的独立同分布样本,提高了相位滤波质量。该方法联合高质量区域的相位信息来拟合低质量区域的干涉相位梯度,相对于传统的局部频率(坡度)估计方法,不但在降低相位噪声方面性能优越,而且还能够在很大程度上保持干涉相位条纹的连续性。2.提出了一种阴影区域干涉相位的填补方法。雷达下视角较大时,在地形复杂区域容易引起大面积的阴影区域,严重影响了相位解缠绕的精度和速度。定义了一种伪相干系数来锐化高相干区域和低相干区域的边界,并利用斜面模型来近似阴影区域的干涉相位,有利于相位解缠绕的进行,同时提高了相位解缠绕的精度。3.提出了一种利用先验知识进行基线估计的方法。基线是XT-InSAR中解缠绕相位向数字高程图进行转换时所需的关键参数,其精度直接决定转换的高程精度。利用地面先验知识和成像系统的参数,可以降低基线估计所需要的条件数。4.提出了一种联合像素多基线处理时利用Lanczos迭代进行降维处理的方法。联合像素处理能够在存在配准误差的情况下,获得较好的相位和高程估计结果,但随着图像数目的增多,其运算量急剧增加。分析了联合像素处理协方差矩阵的构造特点,并根据信号子空间的特性,进行子空间拟合来获取高程估计信息,有效降低了处理时间。5.提出了一种混合基线InSAR构形下进行地面动目标检测的方法。该方法首先利用局部坡度补偿技术使杂波样本满足局部独立同分布;进而利用联合像素处理进行杂波抑制,联合像素处理方法利用当前检测像素和周围像素的信息进行联合处理,使存在配准误差时杂波信息能够得到最大限度的利用,具有良好的杂波抑制性能;利用杂波抑制后的结果,再进行局部相位解缠绕来拟合动目标所在区域的地形高程干涉相位;最后使用幅度和相位双门限恒虚警来降低动目标检测的虚警概率。根据合成孔径雷达干涉测量不同的工作方式,本文由两部分构成:第一部分(第二章~第四章):主要讨论垂直航迹干涉测量的数据处理方法。内容主要包括提高干涉相位图质量的方法,基线估计方法,以及多基线XT-InSAR的降维处理方法。第二部分(第五章):主要讨论利用沿航迹基线进行地面动目标检测的数据处理方法。内容主要包括沿航向动目标检测方法以及混合基线InSAR构型下SAR-GMTI方法。更多还原

【Abstract】 Synthetic Aperture Radar (SAR) Interferometry is an improtant technique in remote sensing field, including cross-track interferometry (XT-InSAR or XTI) and along-track interferometry (AT-InSAR or ATI). It employs pairs of SAR images to be coherently combined, and can retrieve more information compared to a single SAR image. High quality terrain elevation model (DEM) and terrain deformation can be achieved by using the XT-InSAR, and the AT-InSAR technique can be used for ground moving target indication (GMTI) and current measurement. The wide and potential applications of SAR interferometry in military, civil and scientific researches, make SAR interferometry one of the most active fields in radar and remote sensing areas.Because the pair of SAR images for interferometry are obtained from two different view angles separated by a cross-track baseline, the interferometric phase are contaminated by noise due to baseline decorrelation, volume scattering and signal-to-noise ratio (SNR), etc. Therefore, interferometric phase filtering is one of the most improtant procedures in InSAR processing.Furthurmore, the measurement accuracy of InSAR system parameters ususlly cannot meet the practical requirement, and so it must be improved by using signal processing techniqures.In practice, the data processing speed is also an important factor to be considered. Besides the mentioned above, XTI and ATI functions are expected to be integrated into one system in order to meet different applications.The dissertation addresses the problems met in practical data processing, and presents some useful methods for XT-InSAR and AT-InSAR data processing.The main work of this dissertation are summarized as follows:1. In Chapter 2, a local gradient estimation method for obtaining enough independent and identically distributed (i.i.d.) samples in interferometric phase filtering is presented which is based on the least-squares fitting using the total interferometric fringe information. The method can combine the interferometric phases of the regions with high coherence to fit the phase gradients of the low coherence regions. Compared to the conventional local frequency estimation methods which are based on the assumption of plane surface model, the method can estimate the phase gradient of any curved surface. Thus it can suppress the phase noise greatly while maintain the interferometric fringes continuity well, combining with the conventional mean filtering.2. In Chapter 2, an interferometric phase compensation method for shadow region is presented. For a large radar look angle, the shadow phenomenon is inevitable, which will seriously degrade the accuracy and the efficiency of interferometric phase unwrapping. A pseudo-coherence is given to sharpen the edge between the regions of high coherence and low coherence, and then the inclined plane model is used to approximate the interferometric phase in shadow regions, thus facilitating phase unwrapping and improving the accuracy.3. Baseline is one of the key parameters in transforming unwrapped interferometric phase to digital elevation models in XT-InSAR. A baseline estimation method based on prior knowledge and SAR imaging parameters is presented in Chapter 3.4. Joint pixel processing for multibaseline interferometric synthetic aperture radar is robust to SAR image coregistration errors. But with the increased number of SAR images, the computational complexity is increased rapidly. The characteristic of the covariance matrix used in joint pixel processing method is analyzed in Chapter 4, and a reduced dimension method based on Lanczos iteration is presented which can reduce the dimension of the covariance matrix to slightly more than that of the signal subspace, and then the subspace fitting is utilized to achieve the terrain height estimate, thus reducing the processing time greatly.5. For a hybrid baseline InSAR system, a SAR-GMTI method is presented to detect ground moving target. The method first compensates the local terrain slope in order to obtain independent and identically distributed (i.i.d.) clutter samples, and then use the joint pixel processing method to suppress the clutter. The joint pixel processing method uses the current pixel to be detected and its surrounding pixels to jointly suppress clutter, and thus the clutter can still be effectively suppressed when the SAR images are not accurately coregistered. The phase unwrapping is then used to fit the interferometric phase of the region where ground moving targets may be located. Finally, the amplitude constant false alarm rate (CFAR) and phase CFAR are used jointly to decrease the false alarm ratio, and iteration method is also used to improve the relocation accuracy of the detected moving target.According to the classification of synthetic aperture radar interferometry, the dissertation is mainly composed of the following two parts:Part I (Chapter 2~ Chapter 4): The XT-InSAR data processing is discussed in Part I including the method for improving the interferogram quality, the local knowledge based baseline estimation method and the reduced-dimension method for joint-pixel multi-baseline InSAR processing.Part II ( Chapter 5): Gound moving target detection using along track baseline in AT-InSAR is analyzed. The content mainly includes multi-channel SAR-GMTI method with along-track baseline and SAR-GMIT method in hybrid along-/cross- track baseline InSAR formation.更多还原