基于点目标自动提取的机载SAR图像残余运动估计方法  

Estimation of residual motion in airborne SAR images based on automatic point target selection

作　　者：钟雪莲[1,2] 向茂生[1] 岳焕印[3] 郭华东[4] 

机构地区：[1]中国科学院电子学研究所微波成像技术国家级重点实验室,北京100190 [2]中国科学院研究生院,北京100049 [3]中国科技部遥感中心,北京100036 [4]中国科学院对地观测与数字地球科学中心,北京100094

出　　处：《高技术通讯》2011年第11期1157-1163,共7页Chinese High Technology Letters

基　　金：863计划（2007AA120302）和973计划（2009CB72400304）资助项目.

摘　　要：对前期提出的估计单幅机载合成孔径雷达（SAR）图像中的残余运动误差的点目标多斜视方法（MTPT）进行了理论分析，并针对其存在的手工选点工作量大、受噪声影响严重的缺点，对其做了以下改进：在图像中自动选择点目标以减少工作量，并对所测得的相位进行滤波和加权平均以去除噪声，同时采用循环计算的策略来提高算法的适应性。实际的机载SAR数据试验结果显示，即使是对于点目标稀少的地区，改进的MTPT算法也能正确地估计SAR图像中的残余运动误差。同时，该方法在计算精度和稳健性上明显优于加权相位曲率自聚焦（WPCA）方法。The paper gives a theoretic description of the muhisquint technique with point targets （ MTPT）, an algorithm presented in the early stage of the study for estimating the residual motion errors in airborne synthetic aperture radar （SAR） images, and points out that the algorithm is more labor-consuming due to the manual selection of point targets, and more sensitive to noises. Aiming at these two shortcomings, The following are proposed to improve it. Automatic target selection is adopted to reduce the labor, and phase filtering as well as weighted averaging is applied to remove phase noises. Furthermore, The iteration strategy is utilized to improve its adaptability. It is found through the test for the real airborne SAR data that the refined algorithm can estimate the residual motion errors from SAR images correctly, even for areas with few point targets. Meanwhile, its accuracy and robustness are superior to the weighted phase curvature autofocus （WPCA） method.

关 键 词：机载 合成孔径雷达(SAR)图像 残余运动误差 点目标 重轨干涉 

分 类 号：TN958[电子电信—信号与信息处理]