摘要
针对SAR景象匹配基准图制备适应性差及自动化程度低的缺点,提出了一种基于边缘特征的基准图制备方法。首先,组合多种SAR边缘检测算法进行边缘检测;其次,通过对边缘检测点的密度统计分析,得到不同级别的边缘等级图;再次,通过匈牙利法求解边缘点对一一映射的最优指派问题,计算出不同级别的边缘等级图与边缘检测结果图集之间的相似测度值;最后,根据最小相似测度值得到最佳边缘检测级别,并将其对应的边缘检测结果作为边缘特征基准图。实验结果表明,文中方法具有很强的开放性与场景目标适应性,自动化程度高,能有效地提高景象匹配的匹配概率与匹配精度。
An edge-based reference map preparation methodology for SAR scene matching was proposed to re- solve the problem of weak suitability and low automation in reference map preparation. Firstly, edge features were detected by using a group of SAR edge detectors. Secondly, after statistical analysis of edge point density, differ- ent levels of edge density maps were obtained. Thirdly, by calculating optimal assignment problem based on Hungarian algorithm, similarity was measured between edge density maps and edge detection output maps. Final- ly, optimal edge detection output, which served as edge-based reference map preparation in scene matching, was extracted according to the minimum of similarity measurement. The experimental results show that the methodol- ogy possesses better openness and scene suitability, higher automation and wider practical implementations.
出处
《苏州科技学院学报（自然科学版）》
CAS
2015年第4期57-62,共6页
Journal of Suzhou University of Science and Technology （Natural Science Edition）
基金
航天基金资助项目(0747-0540SITC2099-4)
关键词
景象匹配
合成孔径雷达
边缘特征
基准图
最优指派问题
scene matching
synthetic aperture radar
edge feature
reference map
optimal assignment problem
