自学习型图像视觉检测技术在条烟外观检测中的应用
Application of Self-learning Image Vision Inspection Technology in Cigarette Appearance Inspection
王贺伟 1李华 1陈飞程 1梁佳玉1
作者信息
- 1. 广西中烟工业有限责任公司,广西柳州 545006
- 折叠
摘要
针对卷包车间包装工序条烟外观检测中的图像报错率较高,提出了一种基于自学习型数字图像解析算法的条烟检测的改进方案.使用工业相机采集条包外观的良品图像与缺陷图像,通过对图像的二值化,并在检测算法中引入自学习型算法自动区分图像特征点,提高运算效率.改进后检测系统降低了调试难度,提高了现场快速换牌的便捷性,有效提高了卷包车间包装工序条包的检测速度和精度,降低了条包外观检测的报错率.
Abstract
In response to the high error rate in image reporting during the packaging process of cigarette packaging workshops,an improved scheme for cigarette inspection based on a self-learning digital image analysis algorithm is proposed.Use an industrial camera to capture good and defective images of the appearance of the strip packaging.By binarizing the images and introducing self-learning algorithms into the detection algorithm to automatically distinguish image feature points,the computational efficiency is improved.The improved detection system has reduced the difficulty of debugging,improved the convenience of rapid plate replacement on site,effectively improved the detection speed and accuracy of the packaging process in the rolling and packaging workshop,and reduced the error rate of the appearance inspection of the packaging.
关键词
自学习/图像视觉/检测技术/条包外观Key words
self-learning/image vision/inspection technology/cigarette package appearance引用本文复制引用
基金项目
广西中烟工业有限责任公司项目(GXZYZZ2023D001)
出版年
2024
NETL
NSTL
万方数据