摘要
针对在基于图像的高准确度测量领域中同心圆图像检测这一实际问题,分析并实验验证了Hough变换法的缺点和一般圆拟合法的不足,根据系统光轴与同心圆旋转对称轴重合的特点,提出对图像平面进行区域划分再应用圆拟合法检测的方法.建立了基于距离空间聚类和基于随机采样两种区域划分算法,并且给出了圆心初值误差的极限.通过实验仿真,证明了新算法简单高效和可靠,而基于随机采样的算法更适合于实时检测的应用.
Analyzing and validating the defect of Hough transform and circle fitting method contraposing the issue of concentric circles detection in high precision measurement base on image. The detect method is made use of circle fitting after partitioned the region of image plane is presented based on the axis of system and axial symmetry are superposition. Two algorithm for region partition based on distance space clustering and random sample are found. The limit of initial value error is given. The new algorithm is approve of simple,efficiency and credibility through experimentation. And the algorithm based on random sample fit real time application.
出处
《光子学报》
EI
CAS
CSCD
北大核心
2006年第12期1946-1949,共4页
Acta Photonica Sinica
关键词
同心圆检测
圆拟合
区域划分
距离空间聚类
随机采样
Concentric circles detection
Circle fitting
Region partition
Distance space clustering
Random sample
