基于重要性采样的随机Hough变换圆检测算法  

A randomized Hough transform circle detection algorithm using importance sampling technique

作　　者：付琰[1] 卢荣胜[1] 夏瑞雪[1] 王成顺[1] 李琪[1] 

机构地区：[1]合肥工业大学仪器科学与光电工程学院,合肥230009

出　　处：《电子测量技术》2012年第5期45-48,82,共5页Electronic Measurement Technology

摘　　要：提出了一种新的基于重要性采样的随机Hough变换圆检测算法,利用基于圆的几何特性的边界点搜索策略提高采样效率,通过指定重要性采样技术中的目标分布,计算采样参数权重,使得圆识别更加容易和精确;利用基于聚类技术的峰值检测方法,实现多个圆的同时检测。通过将本文提出的新算法与现有的其他两种算法进行实验对比,表明该新算法在检测精度、检测速度、识别可靠性、抗噪能力等方面都有显著性提高。Hough transform is a widely used and effective method for circle detection.A new randomized Hough transform circle detection algorithm using importance sampling is proposed.A search strategy of edge point is used to improve sampling efficiency.By specifying a target distribution and calculating importance weights of sampled parameters,it＇s easier and more accurate to identify circles.In addition,it＇s possible to detect multi-circles simultaneously in the use of peak detection based on clustering technique.Results from applying our new algorithm and other two methods based on HT are shown.The experimental results indicate that the algorithm has high precision,high speed,high identification reliability and strong robustness to noises.

关 键 词：圆检测 随机HOUGH变换 重要性采样 峰值检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]