摘要
针对难以分离多目标匹配中的特征点,研究了一种多目标匹配算法,该算法采用ORB特征点利用Hamming距离进行匹配,采用对称性测试去除误匹配点,然后获取模板在目标图像上的估计中心点,然后利用估计中心点集的密度峰进行中心点的自适应判断,进而分离不同目标的特征点,同时再次去除误匹配点。然后采用RANSAC方法进行变换矩阵的求解,获取所需要的位置角度等信息。实验结果表明该算法能够有效地分离多目标图像中的目标,具有运行速度快,稳定性好的特点。
According to the feature point matching of multiple targets is difficult to separate,we study a multi object matching algorithm,this algorithm uses the ORB feature points and using the Hamming distance matching,remove the false matching points with symmetry test,then obtain the estimation of the template center point in the target image.and use the density peak center adaptive judgment point set center point,feature points and separate different targets,and use the density peak of estimation center point set adaptive judgment the center point of the feature points and separation of different targets.Then,the RANSAC method is used to solve the transformation matrix and obtain the required location,angle and other information.Experimental results show that the algorithm can effectively separate targets from multi target images,and has the characteristics of fast running speed and good stability.
作者
高建哲
吕文阁
GAO Jian-zhe;LV Wen-ge(Faculty of Electromechanical Engineering,Guangdong University of Technology,Guangzhou 51006,China)
出处
《机电工程技术》
2018年第7期90-93,共4页
Mechanical & Electrical Engineering Technology
关键词
ORB
密度峰聚类
多目标
特征点匹配
ORB
density peak clustering
multi-target
feature point matching
