• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2017, Vol. 53 ›› Issue (1): 157-164.doi: 10.3901/JME.2017.01.157

• 数字化设计与制造 • 上一篇    下一篇

扫码分享

基于机器视觉铸件布氏硬度在线检测技术研究*

单忠德, 张飞, 任永新, 张静, 聂军刚   

  1. 机械科学研究总院先进成形技术与装备国家重点实验室 北京 100083
  • 出版日期:2017-01-05 发布日期:2017-01-05
  • 作者简介:单忠德,男,1970年出生,研究员,博士研究生导师。主要研究方向为绿色制造工艺与装备、先进成形制造技术与装备。E-mail:shanzd@cam.com.cn
  • 基金资助:
    * 国家杰出青年科学基金资助项目(51525503); 20160613收到初稿,20161014收到修改稿;

On Line Detection Technology of the Hardness of Cast Iron Parts Based onMachine Vision

SHAN Zhongde, ZHANG Fei, REN Yongxin, ZHANG Jing, NIE Jungang   

  1. State Key Laboratory of Advanced Forming Technology and Equipment, China Academy of Machinery Science Technology, Beijing 100083
  • Online:2017-01-05 Published:2017-01-05

摘要:

为实现铸件布氏硬度的在线检测,应用基于机器视觉的布氏硬度自动测量系统采集压痕图像,研究压痕图像滤波、压痕图像轮廓直径提取、直径标定系数等算法。根据压痕图像的特点,提出基于粒子群的Snake模型压痕轮廓提取算法。引入压痕直径标定系数,解决了视觉测量中的压痕直径像素与压痕物理直径的换算关系,并对直径标定系数进行最小二乘法拟合,提高了测量精度。应用布氏硬度在线测量装置对 180~210 HBW 标准硬度块进行试验测试。试验表明:测量平均误差为 0.72%,测量精度在±2 HBW之间,测量标准差为125 HBW,装置重复性好,精度高,完全能够满足铸件的布氏硬度在线检测要求。

关键词: 机器视觉, 图像处理, 误差分析, 布氏硬度

Abstract: In order to realize online-detection of casting the Brinell hardness, automatic measurement system for the hardness based on machine vision, is applied to collect the Indentation image, to study the algorithm of image filtering, image contour diameter extraction and diameter coefficient calibration. According to the characteristics of the indentation image, a snake model of indentation contour extraction algorithm based on particle swarm is proposed. The indentation diameter calibration coefficient is introduced to solve the conversion relation between the indentation diameter pixels and indentation diameter in the visual measurement, to achieve the least-squares fitting what can improve the measurement accuracy. The online-detection device of the Brinell hardness is used to test 180-210 HBW standard hardness test block. Test shows that the average error is 0.72%, the measurement precision is between ±2 HBW, the standard deviation is 125 HBW, the device has good repeatability and high precision. So the device can meet the online-detection requirement of the hardness of castings.

 

Key words: error analysis, image processing, machine vision, Brinell hardness