基于机器视觉的工件尺寸高精度测量方法  

A Method of High-precision Machine Vision Measurement of Size of Workpiece

作　　者：李钊宝[1] 汪新新[2] 贾茜媛 

机构地区：[1]中国计量学院质量与安全工程学院,杭州310018 [2]中国计量学院计量测试工程学院,杭州310018

出　　处：《计算机测量与控制》2015年第6期1866-1870,共5页Computer Measurement &Control

摘　　要：通过优化照相机的内部参数,运用基于转轴法的外切矩形法对工件进行定位,确定工件像素级的姿态和尺寸;选取物体的ROI后,运用双线性差值算法,实现工件边缘的亚像素提取,通过黑塞范式直线拟合,找到精确的物体边缘;最终实现了运用500万像素的工业照相机,在全景拍摄条件下,测量尺寸在80-150 mm的工件,测量重复性精度可以达到0.015 mm.Workpiece＇ s posture and size in pixel level can be confirmed through optimizing the camera internal parameter, utilizing minimum enclosing rectangle method law based on the revolution axis to carry on the localization to the work piece. After selecting object ROI, the subpixel edge is extracted by the billnearity interpolation algorithm. Hessian normal form line fitting is utilized to find the accurate object edge. Using the industrial camera of 5 million pixels, workpiece with the measurement in 80 mm to 150 mm is finally achieved under the condition of panorama shooting, with measurement repeatability accuracy reaching 0. 015mm.

关 键 词：照相机的内参数 转轴法 外切矩形 双线性差值 尺寸测量 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]