摘要
本文提出了一种自适应聚类结合迭代拟合新的多圆检测算法。使用自适应聚类结合迭代拟合的检测方法增强了算法鲁棒性而且避免边缘曲线连接性的要求,因此本文提出的算法对部分圆以及非连续曲线圆都具有较好的检测效果。自适应聚类操作考虑多点的分布,提高算法速度的同时避免了虚假检测的存在。最后给出的不同方法实验结果性能比较表明本文提出的算法是合理高效的,在计算机视觉领域具有一定的应用前景。
This paper proposes a new method combining adaptive clustering with iterative fitting for multiple circle detection. The use of adaptive clustering and iterative fitting improves the robustness against noise and without connectivity requirement. So the algorithm is effective for extracting part circle and detecting object without good edges. The use of adaptive clustering avoids false circle detection and significantly reduces the computational time. A comparison of the experimental results with other methods demonstrate that this proposed method performs better in computational time in detecting multiple circle and need little known information.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2006年第z2期1183-1184,共2页
Chinese Journal of Scientific Instrument
关键词
多圆检测
机器视觉
聚类
霍夫变换
multiple circle detection machine vision clustering hough transform
