一种圆形瓶口缺陷定位检测方法

摘要

针对机器视觉检测中缺陷面积较大的圆形瓶口定位误差大的问题,分析了瓶口图像各个角度的特征,设计了一种在缺陷中迭代求解瓶口中心的方法.对拟合外边缘的残差进行排序并消除干扰像素,逐步循环迭代,准确定位了瓶口圆心,克服了缺陷部位像素对于圆目标识别的影响,实现了缺陷的实时定位.搭建了试验平台,用500幅不同缺陷的瓶口图像验证,结果表明该方法针对各种样式的缺陷,定位圆精度明显高于参考文献中典型方法,运行时间仅为0.01 s,证明了本方法在瓶口缺陷的实时定位、识别、分析方面具有明显优越性.%In order to solve the problem of big positioning error of circle bottle mouth image with the larger defect area in quality inspection based on machine vision,features of bottle mouth image on various angles are analyzed,and a novel method to iteratively solve the bottle mouth center under defect is designed.By sorting the residuals of fitting outer edges and removing noisy pixels,looping iteration step by step,the center of bottle circle is accurately positioned.The influence of defect zone pixels to circle object recognition is overcome,and the real-time defect positioning is achieved.Experiment platform is established.Five hundreds bottle mouth images with different defects are used for verification.The results show that under all kinds of defects,the circle position precision of this method is apparently higher than the model methods in references,and the performance period is only 0.01 s.It proves that the proposed method has obvious superiority in real-time positioning,recognition and analysis of bottle mouth defect.