摘要
对相关系数的四种计算方法进行了分析 ,对比了这些计算方法的计算量、评估效果和优缺点。考虑到干涉图中各点的信噪比不同 ,将边缘检测中的方向平滑和滤波器设计中的最大误差最小准则的思想引入到求取相关系数中 ,提出了方向加权相关系数的计算方法。对平滑程度较好、信噪比高的数据施加大的权重 ,对受到严重噪声污染、信噪比低的数据施加小的权重 ,可以得到更合理的相关系数。基于方向加权的相关系数进行配准 ,在干涉条纹密集、相关窗口较小时 ,可以抑制干涉条纹边缘的噪声 ,得到更好的配准效果。文中应用ERS 1/2实际数据进行了验证 ,给出了精配准偏移量和配准后干涉图 。
This paper discusses co registration methods based on correlation coefficient in InSAR data processing. It analyzes 4 calculation methods of correlation coefficient. The computation, evaluation effects, the merits and shortcomings of these methods were compared. Regarding the signal noise ratio is not same for all interferometric data,an orient weighting calculation method is proposed. It introduces the orient smooth in edge detection and the minimizing maximum error in to filter design. It exerts heavy weight on high SNR data, and light weight on polluted low SNR data. It would get a more reasonable correlation coefficient. When interferometric stripe is thick and correlation window is small, it would restrain noise at edge of interferometric stripe and obtain better co registration effect using orient weighting correlation coefficient. The paper verifies the algorithm with ERS 1/2 data, and gives offset of fine co registration and interferometric image. The result indicates the algorithm is valid.
出处
《雷达科学与技术》
2004年第2期108-114,共7页
Radar Science and Technology
基金
"十五"预研资助项目 (No.40 2 0 40 3 0 2 )
关键词
干涉合成孔径雷达
数据处理
相关系数
配准
synthetic aperture radar
interferometry
co registration
correlation coefficient
