基于矩阵相似度的InSAR图像配准方法研究

Study on image co-registration based on matrix similarity degree

图像配准是干涉合成孔径雷达成像的关键技术之一,相关系数是InSAR图像配准的重要指标。本文在分析研究常见相似性度量的基础上,研究了相关系数与矩阵相似度度量之间的关系,研究发现,矩阵相似度度量可涵盖多种图像配准数值指标,且其表达形式精炼,几何意义明确,便于把握图像配准的机理。此外,本文提出了一种加权矩阵相似度度量,并通过仿真数据进行了验证。

Image co-registration is a key step in InSAR processing, while the correlation coefficient is an important index of coregistration. This paper discusses correlation coefficient, and studies the relationship of correlation coefficient and matrix similarity degree. The results show that matrix similarity degree is much terser and has a clearer geometric significance than correlation coefficient in image co-registration, and it is also an index of co-registration. Finally, a method computing the orient weighting matrix similarity degree is proposed. The paper verifies the algorithm with stimulant data and the result indicates the algorithm is valid.