布氏硬度自动测量中多圆检测方法

Multi-circle Detection in Automatic Measurement of Brinell Hardness

圆检测应用广泛,是布氏硬度自动测量的关键。针对圆检测中存在圆分裂、多个圆、不完整圆的情况,提出一种基于交叉圆合并、凸包点迭代纯化最小二乘拟合的圆检测方法。首先对图像进行纹理增强并二值化,其次提取有效区域的最小外接矩形,并得到圆弧与最小外接矩形的三个切点,得到初始圆,然后合并有交叉的圆。最后求合并圆的轮廓点与其凸包的交集,进行迭代纯化最小二乘拟合,最终得到亚像素级的圆半径值。最后通过实际应用测试,验证了论文方法的有效性。

Circle detection has wide utilizations and is the key to automatic measurement of Brinell hardness.Aiming at the ex⁃istence of circular split,multiple circles and incomplete circles in circle detection,this paper proposes a circle detection method based on cross-circle merge and iterative least squares fitting for convex-hull points.Firstly,the image is texture-enhanced and bi⁃narized,and then the minimum circumscribed rectangle of the effective region is extracted.Three tangent points of the arc with the minimum circumscribed rectangle are obtained to get preliminary circle,and the intersecting circles are merged.Finally,the com⁃mon points of the contour points of the merged circle and its convex hull is obtained,the radius of the circle at sub-pixel level is ob⁃tained by iterative least squares fitting.The effectiveness of the proposed method is proved by practical application tests.