转子铜排鱼尾槽尺寸的视觉检测评定算法

摘要

A turbo generator's coil strip slot needs to be strictly tested after its milling.While the traditional manual detection can cause missing and false detection,the correct detection rate is low.This paper presents a machine vision measurement algorithm.Firstly,by using the improved Otsu threshold segmentation method based on the Laplace transformation,the algorithm obtains the precise location of edges.Secondly,subpixel edges are obtained with the moving least square (MLS).Finally,experiments on sample coil strip slot test and verify the algorithm.The experimental results indicate that the algorithm proposed in this paper can fulfill coil strip slot online detection needs and acquire a higher accuracy of repeated measurement than the traditional methods,with the repeated measurement accuracy increasing by more than 50％.%汽轮发电机转子铜排完成鱼尾槽铣削后需要进行严格的检测.由于传统的人工检测方法不可避免地带来错检或漏检,导致检测正确率(检正率)低下,本文中采用机器视觉系统对转子铜排两端鱼尾槽进行尺寸检测.首先,利用基于拉普拉斯变换改进的Otsu阈值分割方法,实现了目标边缘轮廓的精确定位;其次,结合使用移动最小二乘法(MLS)实现了亚像素轮廓的提取;最后,使用鱼尾槽样件对上述提出的算法进行了检测验证.实验结果表明,本文提出的方法可以满足铜排鱼尾槽在线检测要求,与传统方法相比,检测重复测量精度更高,重复测量精度平均提高50％以上.