摘要
针对随机Hough变换中采样随机性带来的大量无效采样和积累导致算法效率过低的问题,对随机Hough变换的基本原理进行研究,提出一种改进的随机Hough变换快速圆检测算法。利用图片边缘细化处理减少噪声因素的影响,通过随机采样一点和采用最小二乘法对该点所在的N×N窗口内的连通域进行圆拟合,得出候选圆以提高采样效率,通过缩小限定验证取值区域减少计算量,进行候选圆的证据累积判断。实验结果对比分析表明了该算法具有更快的检测速度和较好的准确性。
Aiming at the problem of low efficiency caused by random sampling and accumulation in randomized Hough transform（RHT）,the basic principle of the random Hough transform was studied,an improved randomized Hough transform based fast circle detection algorithm was proposed.The influence of noise was reduced using the edge refinement of the image.The candidate circle was rounded out by randomly sampling apoint and using the least squares method in the N×N window where the point was located to improve the sampling efficiency.While by reducing and limiting verification region,the amount of computation load for the accumulation of evidence was reduced to determine a candidate circle.Experimental results show that the algorithm has higher detection speed and accuracy.
作者
朱正伟
宋文浩
焦竹青
郭晓
ZHU Zheng-wei,SONG Wen-hao,JIAO Zhu-qing,GUO Xiao(School of Information Science and Engineering,Changzhou University （Wujin Campus）,Changzhou 213164,China)
出处
《计算机工程与设计》
北大核心
2018年第7期1978-1983,共6页
Computer Engineering and Design
基金
国家自然科学基金项目(51307010)
