中国塑料 ›› 2021, Vol. 35 ›› Issue (12): 114-120.DOI: 10.19491/j.issn.1001-9278.2021.12.019

• 标准与测试 • 上一篇    下一篇

基于机器视觉的横拉薄膜破裂图像检测技术

常昊1(), 杨正昊2, 韦有共2, 柴广1, 李岩舟1()   

  1. 1.广西大学机械工程学院,广西 南宁 530000
    2.桂林格莱斯科技有限公司,桂林 541000
  • 收稿日期:2021-05-19 出版日期:2021-12-26 发布日期:2021-12-22
  • 作者简介:常昊(1997—),男,在读硕士研究生,研究方向为机器视觉检测技术,changhao_0217@qq.com
  • 基金资助:
    广西科技重大专项(桂科AA18242011)

Image detection technology of film stretching breakage based on machine vision

CHANG Hao1(), YANG Zhenghao2, WEI Yougong2, CHAI Guang1, LI Yanzhou1()   

  1. 1.School of Mechanical Engineering,Guangxi University,Nanning 530000,China
    2.Guilin GLESI Scientific Technology Co,Ltd,Guilin 541000,China
  • Received:2021-05-19 Online:2021-12-26 Published:2021-12-22
  • Contact: LI Yanzhou E-mail:changhao_0217@qq.com;lyz197916@126.com

摘要:

针对在薄膜加工中恒温箱内温度过高或过低导致薄膜破裂,但人工因恒温箱的高温和结构遮挡难以及时发现薄膜破裂这一问题,结合目前机器视觉的方法,提出一种薄膜破裂图像检测技术,能够监测薄膜的破裂情况,降低生产损耗。在对横拉薄膜破裂区域进行图像处理方法的研究当中,对常见图像的增强、去噪、边缘分割技术分别进行了分析和比较,找出了适用于横拉薄膜破裂图像的预处理、分割方法。通过对薄膜破裂图像的分析与比较中,依据破裂形式将破裂方式分为小范围破裂和纵向破裂两种,并找到了面积、周长区分两种破裂方式的视觉识别特征参数,为调整加工温度参数起到了积极意义。结果表明,所提出的检测技术能够准确判断薄膜破裂形式,准确率为92.7 %,可满足工业生产需求。

关键词: 横拉薄膜, 机器视觉, 缺陷检测, 小范围破裂, 纵向破裂

Abstract:

Aiming at the problem that the film ruptures due to too high or too low temperature in the incubator during film processing without real time detection due to the high temperature and structural obscuration of the incubator, an image detection technology was proposed by a combination with the current machine vision method. This rupture image detection technology can effectively monitor the rupture of the film and reduce the production loss. In the study of the image processing methods for the rupture area of ??the transversely stretched film, the common image enhancement, denoising, and edge segmentation techniques were analyzed comparatively. The preprocessing and segmentation methods suitable for the ruptured image of the transversely stretched film were determined. Through the analysis and comparison of the film rupture images, the rupture modes were divided into the small?scale rupture and longitudinal rupture according to the rupture form, and the visual recognition characteristic parameters for distinguishing the two rupture modes by area and circumference were discovered. These parameters played a positive role in the adjustment of the processing temperature. The experimental results indicated that the proposed detection technology could accurately determine the form of film rupture with an accuracy rate of 92.7 %, which can meet the requirement of industrial production.

Key words: double stretch film, machine vision, defect detection, small area rupture, longitudinal rupture

中图分类号: