CAJ | 학술논문

光学图像和 SAR 图像的配准是实现多源图像融合和复合分析的基础，该文介绍了一种光学图像和 SAR图像精配准的方法。首先采用人工的方式选择3～4个同名控制点，在此基础上实现图像的粗略配准，保证 SAR与光学图像尺度和旋转的归一化；然后在粗配准的光学图像上提取特征点，并分别采用不同的梯度算子提取 SAR和光学图像的梯度强度，基于梯度强度进行归一化互相关，从而得到不同图像上的同名特征点；最后根据提取的同名特征点构建不规则三角网，并对每个三角网构成的小面元进行图像的精确配准，最终实现SAR和光学图像的精配准。通过采用星载图像的精配准处理和检验，证明了该方法的可行性，能获得较好的配准效果。
광학도상화 SAR 도상적배준시실현다원도상융합화복합분석적기출，해문개소료일충광학도상화 SAR도상정배준적방법。수선채용인공적방식선택3～4개동명공제점，재차기출상실현도상적조략배준，보증 SAR여광학도상척도화선전적귀일화；연후재조배준적광학도상상제취특정점，병분별채용불동적제도산자제취 SAR화광학도상적제도강도，기우제도강도진행귀일화호상관，종이득도불동도상상적동명특정점；최후근거제취적동명특정점구건불규칙삼각망，병대매개삼각망구성적소면원진행도상적정학배준，최종실현SAR화광학도상적정배준。통과채용성재도상적정배준처리화검험，증명료해방법적가행성，능획득교호적배준효과。
The registration of SAR and optical remote sensing image is the basise for fusing of multi-source image and comprehensive analysis. In this paper a new fine registration method for SAR and optical image is proposed. Firstly, three to four corresponding points are selected manually to realize a coarse registration that eliminates the differences in scale and rotation. Many characteristic points in the optical image are detected and the corresponding points in SAR image are extracted using normalized gradient correlations based on the different gradients by operators. An irregular triangle network is constructed using these corresponding points and each triangle region is finely registered. Finally SAR image and optical image are finely registered. Experiment and processed results demonstrate the feasibility of this method.