【作者】 张妍；

【导师】 冯大政；

【作者基本信息】 西安电子科技大学 ， 信号与信息处理， 2013， 博士

【摘要】 干涉合成孔径雷达(Interferometric Synthetic Aperture Radar, InSAR)是根据两幅合成孔径雷达图像对应像素点之间的绝对相位差所反映的距离差来获得目标高度的,但在实际处理的过程中,干涉合成孔径雷达的测量相位被限制在了(-π,π]之间,同时由于噪声、欠采样等因素的影响,要恢复出真实的相位就变得非常困难。因此,相位解缠技术作为InSAR信号处理中的关键技术一直是信号处理领域研究的重点与难点,本论文针对干涉合成孔径雷达的相位解缠技术进行重点研究,具体工作概括如下：1)本论文在Goldstein枝切法的基础上提出了一种改进的枝切法,该方法对干涉相位图中的正负残差点进行组合,再以枝切线的长度和作为评价值对干涉相位图中的正负残差点进行重新替换组合,使得整体枝切线的长度和得以不断减小,最终趋向于最小。这种方法有效的解决了相位解缠时在残差点密集区域容易出现“孤岛现象”的问题。2)为了能提高干涉合成孔径雷达信号处理时相位的解缠精度,本论文在相位解缠过程中利用高精度局部频率估计法代替相位差分法,这是因为局部频率估计法能够更加准确的反映实际地形的变化,同时为了抑制残差点处的错误传播至整个干涉相位图,引入了无网格法,通过在干涉相位图中划分局部支持域有效的限制了误差的扩散。3)提出了一种基于改进枝切法与曲面拟合的合成相位解缠方法,这种方法针对干涉相位图中不同质量的区域采用不同的解缠策略。首先利用枝切法准确性高的特点解缠高质量区域的相位,然后利用曲面拟合的方法对低质量区域的相位进行拟合,最后利用遗传算法对拟合函数进行修正来获得最终的解缠相位。这种方法能够很好的解决枝切法容易出现的“孤岛现象"以及避免最小二乘法存在相位坡度欠估计的情况。4)本文论文还基于全局最小均方算法Total Least Mean Square (TLMS)提出了一种稳健的相位解缠方法。首先从理论上证明了该方法的可行性之后,通过若干个含有噪声的仿真数据验证了该方法实现的简便性以及计算的高效性。该方法不仅能够提高解缠的计算效率,而且能够在不增加滤波器的情况下提高干涉相位图的信噪比。更多还原

【Abstract】 Interferometric synthetic aperture radar (InSAR) measures the height of the object based on the distance difference reflected by the absolute phase difference of the corresponding pixels on two pieces of synthetic aperture radar (SAR). But in fact, the phase difference is restricted to (-π,π), and phase unwrapping become so difficulty because of the influence by noise and under sample and so on. Therefore as a key technique in interference synthetic aperture radar (InSAR) signal processing, the phase unwrapping has been the focus and difficulty of the research. This dissertation will focus on the phase unwrapping in interference synthetic aperture radar (InSAR) signal processing, the main research work summarized as follows:1) The branch-cut method is one of the most important approaches for the phase unwrapping, and this dissertation, an improved branch-cut method is proposed based on the Goldstein branch-cut method. The positive and the negative residues in the interference image are combined in this method, then re-combined and replaced using the length of the branch-cut as criteria, so as to reduce the overall length of the branch-cut gradually and overcome the "isolated island phenomenon" in the unwrapping process effectively.2) To improve the processing precision of phase unwrapping in interferometric synthetic aperture radar (InSAR), the phase difference is replaced by local frequency estimation in this dissertation. The local frequency estimation can reflect topographic change more accurately than phase difference. To prevent the residues error spread all over the image, the mesh-less method is also introduced, which can avoid error propagation effectively by dividing local support domain in interferogram.3) A hybrid phase unwrapping method that combines the improved branch-cut method and surface-fitting method is proposed in this dissertation, the method applies different methods in different quality regions to unwrap the phase. It first uses the improved branch-cut method to unwrap the phases in high quality regions for the benefit of high accuracy; then uses surface-fitting method to fit the phases in low quality regions; finally, optimize the coefficients of the surface-fitting function using gene algorithm to obtain the unwrapped phase. The method can avoid the "isolated island phenomenon" of the branch-cut algorithm and the phase gradient error in the least-mean-square algorithm.4) A robust phase unwrapping method that based on total least mean square (TLMS) is proposed in this dissertation. Firstly practicability of the method can be proved in theory, and the computational efficiency also can be illustrated by some simulated data. The proposed algorithm is especially appealing due to its computational efficiency, and can improve the signal to noise ratio (SNR) of Interferogram.更多还原