哨兵雷达卫星TOPS模式干涉处理研究

发布时间：2018-03-08 16:14

本文选题：TOPS　切入点：哨兵　出处：《武汉大学》2016年博士论文　论文类型：学位论文

【摘要】：哨兵卫星是继ERS和ENVISAT卫星之后欧空局又一颗具有划时代意义的C波段SAR遥感卫星。哨兵卫星宽幅成像模式采用TOPS(Terrain Observation by Progressive Scans)技术,卫星在飞行过程中波束沿方位向从后向前摆动,从而获得一个更大范围的成像区域。与ScanSAR模式相比,TOPS模式影像中多普勒模糊、分辨率和信噪比一致,消除了影像扇贝效应,提高了影像质量。本文进一步分别从影像信噪比、方位向模糊以及时问同步方面分析了TOPS模式干涉性能。TOPS模式每组Burst持续时间较长,提高了雷达回波利用率,但是多普勒中心频率变化达5KHz,干涉处理时配准精度要求达到0.001个像素,否则干涉图相位出现较大的偏差,相邻Burst干涉图相位将出现跳变现象。本文则针对哨兵卫星TOPS模式影像干涉处理理论和方法做全面的论述。鉴于哨兵卫星影像质量和卫星辅助参数均达到了很高的精度,论文优先考虑基于影像几何的初配准。但是该方法依赖于精密轨道。考虑实时处理的需要,论文在影像几何配准的基础上,采用影像相关配准算法矫正较大的方位向系统配准偏差。考虑到重复观测过程中地物位置和物理特征的变化导致的影像的失相干问题,本文增加了点目标配准的方法以增加相关性配准的可靠性,最后利用Burst模式的重叠区域进行增强谱分集配准,优化方位向系统性配准误差。哨兵影像精配准处理依赖于增强谱分集方法,但是该方法易受到方位向形变、电离层闪烁等因素的干扰,其差分相位不但包含配准误差信息,还存在方位向形变等其他信号。本文则讨论了增强配准的干扰原理以及相应的干扰排除方法。哨兵卫星重访周期更短,理论上可以监测更大速率的沉降。本文则采用StaMPS软件对哨兵数据进行PSInSAR处理。然而StaMPS技术在进行大气效应分离时,并没有像经典PS技术一样采用先验模型提前分离出主要的形变信息。当沉降速率较大时,PS点之间所含信息成分差异较大,传统的经典滤波器则无法保证这些信息的正确分离,本文提出平滑样条滤波可以提高信息分离的精度,并可抗拒解缠错误的干扰。哨兵卫星为我们利用重叠区域检验时序分析结果提供了可能。哨兵卫星影像中所有组Burst参考同一轨道,且相邻Burst成像时间仅差数秒,理论上影像重叠区域时序处理结果应该是一致的。本文基于墨西哥地区和天津地区数据对每组Burst进行单独处理,然后根据方位向重叠区域评估PS处理结果的精度。
[Abstract]:Sentinel satellite is another epoch-making C-band SAR remote sensing satellite after ERS and ENVISAT satellites. The wide-amplitude imaging mode of Sentinel satellite is based on TOPS(Terrain Observation by Progressive Scans. during flight, the beam is swinging back and forth along the azimuth. Compared with the ScanSAR mode, the Doppler blur, the same resolution and the same signal-to-noise ratio can eliminate the scallop effect and improve the image quality. In the aspect of azimuth fuzziness and timely synchronization, the interference performance of TOPS mode. The Burst duration of each Burst group of trips mode is longer, which improves the radar echo utilization ratio. However, the Doppler center frequency changes to 5 KHz, and the registration accuracy is 0.001 pixels in the interferometric processing. Otherwise, the phase of the interferogram will deviate greatly. The phase jump of adjacent Burst interferogram will occur. In this paper, the theory and method of Sentinel satellite TOPS mode image interferometry are discussed. In view of the high precision of satellite image quality and satellite auxiliary parameters, The paper gives priority to the initial registration based on image geometry, but the method depends on the precision orbit. Considering the need of real-time processing, the paper is based on the image geometry registration. The algorithm of image correlation registration is used to correct the large registration deviation of azimuth system. Considering the problem of image incoherence caused by the change of ground object position and physical characteristics during repeated observation, In this paper, the method of point target registration is added to increase the reliability of correlation registration. Finally, enhanced spectral diversity registration is carried out by using overlapping region of Burst model. Optimization of azimuth systematic registration error. Sentinel image registration process depends on enhanced spectral diversity method, but this method is easily disturbed by azimuth deformation and ionospheric scintillation, and its differential phase not only contains registration error information. There are also other signals such as azimuth deformation. In this paper, the interference principle of enhanced registration and the corresponding interference elimination method are discussed. The re-visit period of Sentinel satellite is shorter than that of Sentinel satellite. In this paper, StaMPS software is used to process Sentinel data by PSInSAR. However, when StaMPS technology is used to separate atmospheric effects, The prior model is not used to separate out the main deformation information in advance as the classical PS technique. When the settlement rate is large, the difference of information components between PS points is large, but the traditional classical filter can not guarantee the correct separation of these information. In this paper, it is proposed that smooth spline filtering can improve the accuracy of information separation. The Sentinel satellite makes it possible for us to use the overlapping region to test the time sequence analysis results. All groups of Burst in the Sentinel satellite image refer to the same orbit, and the imaging time of adjacent Burst is only several seconds. In this paper, each group of Burst is processed separately based on the data of Mexico and Tianjin, and then the accuracy of PS processing is evaluated according to the azimuth overlapping region.
【学位授予单位】：武汉大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：P236

【相似文献】