摘要
针对检测图像多圆问题,该文提出一种利用圆的对称性与旋转变换相结合的圆检测新方法。从圆的几何性质出发,在图像边缘点上以过圆心且平行横坐标的直线为对称轴搜索对称点集P1;然后图像做旋转变换,再次对称点搜索并做反旋转变换得到点集P2;利用圆的旋转对称性,圆上的点集即是点集P1与P2的交集,从而实现圆的检测。实验结果表明,该算法与当前应用最广的RHT、RCD圆检测方法相比具有检测速度快,实现简单等优点。
Aiming at the detection of Multi-Circle,a novel method based on a combination of circular symmetry and rotation transformation is proposed.According to the circle geometrical properties,taking the straight line as axis of symmetry through the center of circle and paralleling to abscissa,symmetric points named point set P1 are searched from the points on the edge of the image.After the image rotated,symmetric points are searched from it,with reverse rotation transformation,and named point set P2.Making use of circle rotational symmetry,the point set that is the intersection of P1 and P2 is on a circle,which will realize the detection of circle.Compared with Randomized Hough Transform(RHT) and Randomized Circle Detection(RCD) which are most widely applied,the experimental results show that the algorithm has the advantages of rapid detection and simple implementation.
出处
《光电工程》
CAS
CSCD
北大核心
2011年第9期137-141,共5页
Opto-Electronic Engineering
关键词
圆检测
对称性
旋转不变性
计算机视觉
multi-circle detection
symmetrical characteristic
rotation invariant
computer vision