【作者】 谢先明；

【导师】 皮亦鸣；

【作者基本信息】 电子科技大学 ， 信号与信息处理， 2011， 博士

【摘要】 干涉合成孔径雷达(InSAR)具有全天候、全天时、高效率获取目标区域数字高程地图(DEM)的能力,已广泛地应用在地形测绘、地表形变监测、目标探测等领域。多基线InSAR技术是InSAR技术的扩展,克服了传统InSAR技术对系统噪声和大气效应敏感等固有的缺陷,具有获取高精度DEM数据的能力,已成为InSAR技术应用研究的热点。本文对InSAR数据处理的主要环节,包括SAR图像配准、去平地效应等进行了分析与讨论,重点对干涉图滤波及缠绕相位展开等InSAR关键技术进行了较为系统、深入的研究。此外,重点对多基线InSAR关键技术—多基线干涉相位估计技术进行了较为深入的研究。本文主要工作和贡献如下:(1)对SAR图像配准、平地效应去除、相位噪声抑制等InSAR数据处理的主要环节开展了较为深入的研究。针对常用滤波算法难以在有效滤除噪声的同时保持干涉图条纹完整性的问题,提出了一种组合滤波方法;利用Goldstein滤波和均值滤波各自的特点,在滤除相位噪声的同时很好地保持了干涉图条纹的边缘特性。(2)对常用的相位展开方法进行了较为系统、深入的研究。针对InSAR技术应用中的难点—条纹密集干涉图和复杂地形干涉图的展开问题,把传统路径跟踪策略与Unscented Kalman filter(UKF)结合起来,提出了一种基于路径跟踪策略的UKF相位展开方法;实验结果表明该方法有效地展开了条纹密集且复杂的干涉图,且与传统方法相比具有较高的精度和较强的稳健性。在此基础之上,把人工智能的搜索策略引入到UKF相位展开算法之中,提出一种基于人工智能搜索策略的UKF相位展开方法;利用人工智能的搜索策略在实现传统路径跟踪策略的同时,又更加充分的利用了相邻已展开像元的信息,进一步提高了相位展开精度。(3)针对相位展开问题的非线性和非高斯特性,提出一种不受模型噪声统计特性和问题的非线性影响的粒子滤波(PF)相位展开方法;实验结果表明该方法具有较强的稳健性。(4)对经典多基线相位估计方法进行了较为系统的研究。针对现有多基线相位估计方法通常存在算法适应性和稳健性不强的问题,把UKF应用到多基线干涉相位估计之中,提出一种多基线UKF相位估计方法,并通过实验验证了该方法的有效性和稳健性。此外,利用扩展粒子滤波(EPF)的强大数据融合能力与不受模型噪声特性影响的特点,提出一种多基线EPF相位估计方法,并在多基线相位展开实验中获得了较好的效果。(5)提出一种适用于三基线以上的干涉相位估计方法;通过选取适当的基线组合,在展开最短基线干涉相位基础上,利用最大似然频率估计器提取每一复像元随基线变化的频率,有效地实现了长基线干涉相位估计。更多还原

【Abstract】 Interferometry Synthetic Aperture Radar (InSAR) is a novel remote sensing technique that has been developed recently. Due to its capability of all-weather and all-time, and high-efficiently acquiring digital elevation maps (DEMs) of target scene, InSAR has been applied widely in topography mapping, the detection of surface deformations, the detection of target etc. It is well-known that conventional InSAR is sensitive to system noise and atmospheric effect. Multi-baseline InSAR, considered as the extension of conventional InSAR, is able to acquire high-precision DEMs of earth surface and has become an issue of remote sensing technology application research since it can overcome inherent defect of conventional InSAR.The principle and the methods of main steps of InSAR data processing, including SAR image registration, flat earth removal, noise filtering, phase unwrapping etc, are investigated and discussed, and two key steps in InSAR data processing, including noise filtering and phase unwrapping are studied in detail. In addition, multi-baseline phase unwrapping technique, considered as the key technology of multi-baseline InSAR data processing, is also studies in detail.The main work and innovations accomplished in this dissertation are as follows:(1) The techniques of image registration, flat earth removal and noise filtering etc, is investigated and discussed in this dissertation. It is difficult to normal filtering algorithms to remove effectively phase noise in interferogram, and with good edge preservation. To solve above problem, a combined filtering algorithm that can remove phase noise effectively, and with good edge preservation is proposed in this dissertation.(2) Conventional phase unwrapping algorithms are investigated and summarized. As is known it is difficult for conventional algorithms to unwrap effectively the interferogram with complex fringe, or with dense fringe. To solve above problem, a novel phase unwrapping algorithm based on the unscented Kalman filter (UKF) is proposed. This method is the result of combining an UKF with the path-following strategy and an omni-directional local phase slope estimator. Simulation and real data processing results validate the effectiveness of proposed method, and show a significant improvement with respect to some conventional phase unwrapping algorithms in some situations. On this basis, combining an UKF with the artificial-intelligence search strategy,an novel UKF phase unwrapping algorithm based on the artificial-intelligence search strategy is proposed. The artificial-intelligence search strategy will ensure that this algorithm performs noise filtering and phase unwrapping along the path from high-quality region to low-quality region, and will guarantee unwrapped neighbors information are fully exploited. Therefore, phase wrapping accuracy can be improved further.(3) As for the nonlinearity and the non-Gaussian characteristic of the phase unwrapping problem, a novel phase unwrapping algorithm based on the particle filter (PF) is proposed. This method provides independence from noise statistics and is not constrained by the nonlinearity of the problem. Simulation and real data processing results validate the effectiveness of proposed method.(4) Classical multi-baseline phase unwrapping algorithms are investigated and summarized. As is well-known that adaptability and stability of conventional algorithm isn’t strong enough. To solve above problem, a novel multi-baseline phase unwrapping algorithm based on the UKF is proposed. The performance of the proposed method from synthetic data is illustrated. In addition, a novel multi-baseline phase unwrapping algorithm is proposed by using the strong data fusion capacity of extended particle filter (EPF), and the effectiveness of proposed method is validated by Simulation data processing results.(5) A novel multi-baseline phase estimate algorithm, applied in SAR interferometry with more than three baselines, is proposed. This method consists of two steps: firstly, selecting the appropriate set of baselines and unwraping the interferogram associated with the shortest baseline; then gaining the unwrapped phase of the longest baseline by using the maximum likelihood estimator to extract the frequency of any complex pixel.更多还原