摘要
为解决物体产生局部变形之后标志点的匹配问题,提出了一种面向稀疏标志点的特征描述算法。该算法受点特征直方图算法启发,利用标志点与邻域点欧式距离和角度关系,实现了当前标志点的唯一描述,从而建立变形前后标志点的对应关系,并能筛选出几乎不发生形变的点对。实验统计数据表明,该算法求解出的对应关系正确率接近100%,利用算法结果匹配后的平均偏差在0.03 mm以内,具有较强的稳定性。
A feature description algorithm for sparse marked points is proposed,which is used to solve the matching problem of non-coding marked points after local deformation.Based on the point feature histogram(PFH)descriptors,the euclidean distance and angle relationship between the marked point and the neighborhood points is used in this algorithm to implement the unique description of the current marker point,thus establishing the corresponding relationship between the marker points before and after the deformation,and determining the change of each point before and after deformation.According to the experimental statistics,the correction rate of the corresponding relationship approaches to 100%,and the average error after matching by the description is less than 0.03 mm.Therefore,this algorithm has strong stability.
作者
石诚
程筱胜
崔海华
方舟
张逸
胡广露
SHI Cheng;CHENG Xiaosheng;CUI Haihua;FANG Zhou;ZHANG Yi;HU Guanglu(School of Mechanical and Electrical Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)
出处
《机械制造与自动化》
2020年第5期200-202,206,共4页
Machine Building & Automation
关键词
机器视觉
标志点
特征描述
匹配
machine vision
marked point
feature description
matching
