结合边缘和统计特征的末制导SAR图像匹配

提出一种用于末制导的合成孔径雷达（synthetic aperture radar, SAR）图像实时匹配方法。考虑参考图与实时图尺度、方向等差异，制备多幅参考图像，基于最小错误概率准则，对每幅参考图像提取边缘，由边缘图形成用于匹配的包含边缘两侧地物区域信息的参考模板。提出一种利用边缘信息和SAR图像统计特性的相似性度量准则。采用金字塔分层结构的搜索策略，提高了匹配速度。该方法综合利用了边缘和灰度统计信息，避免了对实时图的边缘提取。实验验证了算法的有效性。

SAR image matching using both edge and statistical characteristic for terminal guidance

A real time method of matching synthetic aperture radar (SAR) images for terminal guidance is presented.Considering the difference of resolution and orientation between reference and sensed images,multiple reference images are prepared.Based on the minimum error probability criterion,edges in each reference image are extracted.Reference templates formed from the edge images are used for real time matching,which contain the area information of the objects on both sides of the edges.A new criterior of similarity measure based on edge information and the statistical characteristic of SAR images is presented.The matching is performed in a series of images depicting the same scene,but in different resolution,i.e.,in a resolution pyramid.Using this hierarchical structure reduces the computational load significantly.This method uses both the edge and statistical characteristic information and avoids edges extraction in the sensed image.The experiment result shows that the method is effective for use in synthetic aperture radar terminal guidance.