【作者】 李燕平；

【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(SAR)具有全天候、全天时和远距离成像的特点,可以大大提高雷达的信息获取能力,特别是战场感知能力,对军用和民用均有重要的应用价值。SAR是通过采用宽频带信号来实现距离维的高分辨率,通过载体的运动形成长的线性合成阵列来实现方位维的高分辨率。近十余年来,SAR已经在各种平台上得到了广泛应用,其载体可能是中、低空飞行的轻型飞机、直升机或者无人机等。受气流扰动的影响,载体在飞行过程中不可避免地会偏离理想的匀速直线运动状态产生运动误差,导致成像质量严重下降。特别是在目前国产惯导精度有限的情况下,研究基于回波数据的运动补偿方法对实现机载SAR的高分辨成像具有重要意义。常规的运动补偿方法主要针对正侧视的窄波束SAR,并且在处理中忽略了地面的地形起伏。然而在许多实际应用中这些条件并不满足,比如雷达可能工作在斜视模式,雷达的波束宽度可能较宽,地面的地形起伏也可能较大,因而有必要对常规运动补偿方法在各种情况下的适用性进行深入研究。与传统的单基SAR相比,收发分置的双基SAR具有很好的技术优势,比如作用距离更远、获取信息更丰富、机动性和隐蔽性更高、抗干扰和抗截获性能更好等。这些优势使得双基SAR在军事应用、资源调查、InSAR三维重建、地壳形变监测等方面有着广阔的应用前景。然而这些好处的获得是以系统复杂性为代价的,双基SAR还需要突破成像算法和运动补偿等关键技术。本论文主要围绕以上几个方面做了一些工作,现概括如下:1.系统地研究了机载正侧视SAR的运动补偿。在分析正侧视SAR运动误差模型的基础上,提出了一种基于回波数据的运动参数提取和运动补偿方案,有效降低了系统对惯导的要求。该方法的基本思路如下:首先对回波数据沿方位向划分子孔径,接着对子孔径多个距离单元的数据估计多普勒调频率,然后利用多普勒调频率对运动参数进行加权最小二乘估计,最后利用估计的运动参数分别完成视线方向和沿航向运动误差的补偿。2.对运动补偿中的几何形变问题进行了分析,提出了校正方法。3.分析指出了地形起伏较大、雷达波束宽度较宽及雷达斜视角较大时常规运动补偿方法的局限性,分别给出了这三种情况下的改进方法。同时也分析了运动补偿后的残余误差对最终SAR图像的影响。4.分析了相位梯度自聚焦补偿算法中特显点选取原则的不足,提出了一种改进算法。该算法可以自适应地选取SAR图像中质量最好的孤立特显点用于自聚焦,在存在连片强杂波的场景中自聚焦效果更好。5.提出了一种适用于平行等速双基SAR的扩展Chirp Scaling成像算法。该算法通过级数反演法来计算目标的二维频谱,通过双基参数的数值近似来得到距离空变量的解析表达式,通过CS变标的方法来实现目标的距离徙动校正。该算法可看作是单基SAR中的ECS算法在双基SAR中的推广,其优点在于不需要插值、适用范围广、精度高。6.提出了一种适用于小斜视角双基聚束式SAR的频率变标算法。该算法可看作单基SAR的频率变标算法在双基情况的推广,能在长基线情况下实现高分辨成像。另外,该算法将数值方法和方位Deramp技术相结合,解决了星载双基聚束情况可能出现的方位频谱混迭的问题。7.针对斜视角较大的情况,应用非线性变标的思想,提出了一种适用于大斜视角双基SAR的频率变标算法。该算法进一步考虑了随距离变化的二次距离压缩,可以用于较大斜视角的双基SAR成像。8.研究了卫星/飞机组合的这种广义双基SAR的成像算法问题。分析指出驻相点展开法的精度与收发平台的速度及收发系统的斜视角度有关,因而该方法不适合星机双基SAR。提出了两种计算星机双基SAR目标二维频谱的方法,第一种为解析式法,该方法的运算量小,精度较高,适合方位波束宽度较窄的星机双基SAR;第二种为数值方法,该方法的运算量比第一种大,但优点是可以得到准确的目标二维频谱,适合各种情况的星机双基SAR。9.研究了双基SAR的运动补偿。通过对双基SAR运动误差模型的分析,提出了一种基于数据的收发平台三维运动误差的估计和补偿方法。该方法直接从回波数据中估计收发平台运动误差的合成值以用于运动补偿,利用构造的相位补偿函数来校正收发平台前向速度波动造成的运动误差。更多还原

【Abstract】 Synthetic Aperture Radar (SAR), which has the characteristic of all-weather, day/night and long range, can enhance radar’s information acquisition capability, especially the battlefield awareness ability, and has great value in both civilian and military applications.In SAR, high geometric resolution in range is obtained via transmission of large bandwidth pulses. On the other hand, high resolution in azimuth is the result of an intensive coherent data processing operation aimed at synthesizing an antenna array. In recent years, SAR is widely used in various platforms, such as the minitype airplane, helicopter and Unmanned Aerial Vehicle. Due to atmospheric turbulence, the SAR platform trajectory is often subject to small deviations from the ideal straight line, resulting in an incorrect interpretation of the geometry of the SAR image or a loss of focus in the image. To account for such errors, flight parameters are measured onboard with Global Positioning Systems (GPS) and Inertial Navigation Units (INS). However, the accuracy of GPS and INS available in our country is relatively low. Therefore, the research on raw data based motion compensation is of great importance for high resolution airborne SAR imaging.The conventional motion compensation method is mainly applicable for the broad-side imaging case, and assume that the beamwidth in azimuth is narrow enough. In addition, topographic variations within the scene is neglected during motion compensation. However, in many applications, these assumptions can not be satisfied. For example, the SAR system may be operated in the squnit mode, the beamwidth of the radar may be wide and the topographic variations within the scene may be severe. So it is worth analysising the limitations of the conventional motion compensation method.Compared with the conventional monostatic SAR, the bistatic SAR uses a separate transmitter and receiver, flying on different platforms, which enables the exploitation of additional information contained in the bistatic reflectivity of targets. The bistatic SAR also has many other advantages like: flexibility, long range, reduced vulnerability for military applications, ability to use multilevel interferometry, etc. However, these increased advantages are paid for by an increased complexity in designing bistatic SAR systems. Besides technical problems such as the synchronisation of the oscillators, the bistatic imaging algorithm and motion compensation method have still not been sufficiently resolved.The primary contributions of this dissertation, which is devoted to the above aspects, are summarized as below:1. The motion compensation method for airborne broad-side looking SAR is studied. Based on the analysis of the motion error model of broad-side looking SAR, a novel method for raw data based motion parameters extraction and motion compensation is presented, which can reduce the requirement for the INS. This method contains the following procedures: first, the raw data is divided along the azimuth direction. Second, the Doppler rate of each range bin of the sub-aperture data is estimated. Third, the motion parameters of the aircraft are estimated by the Weighted Least-Square method. Finally, the motion errors in the line of sight direction and the aizmuth direction are corrected respectively using the estimated parameters.2. The geometric distortion problem in motion compensation is analysed, then a method is presented for geometric distortion correction.3. The limitations of the conventional motion compensation method in the case of topographic variations, wide beamwidth and large squint angle are analysed. Then, the improved methods are presented for these cases. Also, the effect of the remaining error after motion compensation on the final SAR image is analysed.4. The limitation of the dominant scatter selection method in phase gradient algorithm (PGA) algorithm is analysed, then an improved method is presented. This method can adaptively select the best isolated scatters for the phase error estimation. It is shown that the improved PGA algorithm outperforms the original one especially for the scene in which the bright scatters are not isolated.5. An extended chirp scaling (ECS) algorithm is presented for the bistatic spatial invariant configuration based on series reversion and numerical computation. In this method, the two dimentional point target spectrum is first computated by series reversion. Then, an analytic expression of the range migration factor is obtained by numerical approximation of the bistatic parameters. Finally, the range migration correction is implemented by the Chirp Scaling operation. This algorithm can be considered as an extension of the monostatic ECS algorithm to the bistatic case. The proposed algorithm has the advantages of wide application scope, high accuracy and relativly low computation load, since no interpolation is required during the whole processing chain.6. Based on the analysis of the signal model, a novel frequency scaling algorithm (FSA) for the azimuth-invariant bistatic SAR is presented. This algorithm can be considered as an extension of the monostatic FSA algorithm to the bistatic case. High resolution imaging can be achieved by using this method even in the long baseline case. In addition, this algorithm combine the numerical computation method with the Deramp technique, thus the so-called azimuth spectral folding effect in the spaceborne case can be overcomed.7. Based on the idea of nonlinear chirp scaling, an improved FSA is presented for the bistatic SAR with high squint angles. This algorithm compensates the range-variant secondary range compression factor, thus improving the focus ability of the algorithm in the high squint angle case.8. The general bistatic SAR imaging method is investigated, especially for the Hybrid spaceborne/airborne bistatic SAR. It is pointed out that the accuracy of the point target spectrum derived by Taylor series expansion around the point of stationary phase is determined by both the velocities and the squint angles of the transmitter and receiver. As the satellite has a larger velocity than the airborne platform, and there may be a large difference between the transmitter’s and the receiver’s squint angle, the point target spectrum derived by this method is not accurate sufficiently. Then, based on the method of stationary phase and the bistatic geometry, the two methods are proposed to compute the point target spectrum of the hybrid spaceborne /airborne bistatic SAR. One is based on analytical expression computation, while the other is based on numerical computation. The first method has the advantage of high accuracy and low computation load, while the second method is totally accurate and has a relatively larger computational load.9. The motion compensation method for the bistatic SAR is studied. Based on the motion error model of the bistatic SAR, a method for estimating and compensating the three dimentional motion errors of the the transmitter and receiver is presented. In this method, the sum of the motion errors of the transmitter and receiver is directly estimated from the raw data. Then, these estimated parameters is used for motion compensation. The forward velocity variations of the transmitter and receiver is compensated by using a phase compensation function.更多还原