一种随机Hough变换检测圆的优化方法

Optimization Method of Randomized Hough Transform to Detect Circle

随机Hough变换是常用的图像测圆方法,当图像数据杂乱时,随机Hough变换的结果不理想且检测实时性差。针对激光扫描检测直缝焊管焊缝噘嘴问题,提出了一种优化的随机Hough变换检测圆方法。首先计算激光扫描所得轮廓离散点的曲率值,然后采用K均值聚类法从轮廓图像中分离出圆弧数据点,最后使用随机Hough算法检测圆。实验表明,本文方法可以准确而快速地计算出焊管径向横截面二维轮廓圆的圆心和半径,可以满足工业实际应用需求。

The randomized Hough transform is commonly used in the image processing,but when the data is cluttered,the results of the randomized Hough transform are not very good,and the detection speed is reduced.An optimization method of the randomized Hough algorithm is presented and this improvement is aimed at the weld of the pout detection of straight seam welded pipe.First,the curvature of the discrete points is calculated,the method of K-means clustering is used to separate the arc points from the whole data,and then the randomized Hough transform is used to detect circle by using these points.The experiment proves that it can accurately and quickly calculate the circle＇s center and radius of the radial cross section＇s 2D profiles of the welding pipes,and fully conform to the requirements of industrial application.