基于角点响应函数的加权Hausdorff距离匹配算法  

Weighted Hausdorff Distance Matching Algorithm Based on Corner Response Function

作　　者：徐一鸣[1] 刘晓利[1] 刘怡昕[2] 

机构地区：[1]瞬态物理国家重点实验室,南京理工大学,江苏南京210094 [2]南京炮兵学院,江苏南京211132

出　　处：《探测与控制学报》2009年第6期29-32,37,共5页Journal of Detection & Control

摘　　要：常见的图像匹配算法对目标遮挡以及非零均值噪声比较敏感,为此提出一种基于特征点的图像匹配算法,根据SUSAN算子提取图像的角点作为特征点,然后采用加权的部分平均Hausdorff距离对两个角点集合进行计算,通过引入角点响应函数的差异来抑制接近中心的误匹配点对图像匹配造成的干扰,从而实现较恶劣环境下目标的配准。实验结果证明了该算法具有较强的鲁棒性。Common image matching algorithms are sensitive to target occlusion and noise of non-zero mean. To resolve this problem, a feature point based matching algorithm is proposed. Corners of template and input images found by using SUSAN operator, and the similarity of the two corner sets is calculated with weighted partial mean Hausdorff distance. The algorithm uses the difference between corner response functions to suppress the interference from error corners which are near to the matching center, and realize the image registration under bad conditions. The experiment has proved the robustness of this algorithm.

关 键 词：图像配准 特征点匹配 SUSAN角点检测 加权HAUSDORFF距离 角点响应函数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]