文档介绍：
一种改进的基于边缘测度的
加权 Hausdorff 距离景象匹配算法#
唐宇,凌志刚**
5
10
15
20
25
30
35
(湖南大学电气与信息工程学院)
摘要:目前,Hausdorff 距离在图像匹配中得到了广泛地应用。但 Hausdorff 距离原型算法采
用的最大最小距离对出格点噪声和遮挡非常敏感,即使两幅图像非常相似,少数出格点噪声
的存在也可能导致距离计算的严重错误。据此,本文提出了一种改进的基于边缘测度的加权
Hausdorff 距离算法(EM-WHD),该算法集合结构张量和局部均值比方法构造出边缘测度
图像,并据此提取图像边缘,为提高算法对各种干扰的鲁棒性,利用边缘测度值来估计对应
边缘之间的边缘响应,在计算加权 Hausdorff 距离时考虑了几何信息和边缘测度信息。实验
结果表明,该算法对噪声、旋转和遮挡等干扰有较强的鲁棒性。
关键词:加权 Hausdorff 距离;边缘测度;景象匹配辅助导航;鲁棒性
中图分类号:请查阅《中国图书馆分类法》
An Improved Scene Matching Algorithm with Weighted
Hausdorff Distance Based on Edge Measure
Tang Yu, Ling Zhigang
(School of Electrical and Information Engineering,Hunan University)
Abstract: Currently, Hausdorff distance has been widely used in image the
minimum and maximum distance that hausdorff distance prototype algorithm used is very
sensitive to noise and occlusion of outliers,even if the two images are very similar, few outliers
presence of noise may also cause the distance calculation serious ,an improved
algorithm with weighted Hausdorff distance based on the edge measure (EM-WHD) is
a reference image and a sensed image, edge measure function is constructed based
on the structure tensor method and ratio of average improve the robustness of
algorithms for various disturbances ,use the values of edge measure to estimate the response
between corresponding edges, geometric information and edge measure information is taken into
account in the calculation of the weighted Hausdorff results show that,the
algorithm is robust to noise,rotation and occlusion.
Key words: Weighted Hausdorff distance ; Edge measure; Scene matching aided navigation;
Robust
0 引言
景象匹配由于具备定位精度高、自主性好等优点而受到人们的高度重视,并
在无人飞行器中得到了广泛应用。无人机景象匹配辅助导航[1-4]技术利用机载高
分辨率成像雷达或光电图像传感器实时获取地面景物图像,并利用图像匹配技术
确定当前获取图像在机载计算机中预先存储的地物景象数字参考图中的位置,进
而精确地计算出飞行器位置,以修正航向误差同时消除惯导系统长时间工作的累
基金项目:高等学校博士学科点专项科研基金(201**********)
作者简介:唐宇(1989-),男,硕士研究生,研究方向:图像处理,机器视觉
通信联系