吉林大学学报(信息科学版) ›› 2016, Vol. 34 ›› Issue (3): 413-418.

• 论文 • 上一篇    下一篇

铝型材表面喷涂质量的图像检测与研究

岳晓峰, 胡继文   

  1. 长春工业大学机电工程学院, 长春130012
  • 收稿日期:2015-07-29 出版日期:2016-05-25 发布日期:2016-12-21
  • 作者简介:岳晓峰(1971—), 男, 吉林通化人, 长春工业大学教授, 博士, 硕士生导师, 主要从事机器视觉及智能检测研究(Tel) 86-13844086185(E-mail)yuexiaofeng@ ccut. edu. cn; 通讯作者: 胡继文(1988—), 男, 河南信阳人, 长春工业大学硕士研究生, 主要从事机器视觉及智能检测研究, (Tel)86-13756986386(E-mail)314535209@ qq. com。
  • 基金资助:

    吉林省科技厅基金资助项目(20110303)

Image Inspection and Research of Aluminum Surface Coating Quality

YUE Xiaofeng, HU Jiwen   

  1. College of Mechanical and Electrical Engineering, Changchun University of Technology, Changchun 130012, China
  • Received:2015-07-29 Online:2016-05-25 Published:2016-12-21

摘要:

针对用人眼检测铝型材表面喷涂质量效率低且容易疲劳的问题, 提出了基于数学形态学颗粒分析和改进后的模糊核聚类的图像检测方案。对铝型材表面喷涂的图像进行形态学颗粒分析, 利用数学形态学的并行结构特性通过核函数在特征空间中进行模糊核聚类, 从而达到质量检测的目的。实验结果表明, 该方法收敛速度达到18. 4 s, 分类准确率达到91. 6%, 在整体性能上超过传统的聚类方法, 并具有较强的鲁棒性。

关键词: 数学形态学, 颗粒分析, 核函数, 聚类分析

Abstract:

Aluminum surface coating quality in production enterprises generally inspected by human eyes.Because the human eye can easily fatigue, and detection is inefficient. In view of this kind of situation, we proposed the method of image inspection based on the particle analysis of mathematical morphology and fuzzy
kernel clustering. Firstly we use morphology particle analysis for the image coating surface of the aluminum, then se parallel structure characteristic of the mathematical morphology by introducing kernel function to fuzzy kernel lustering in the feature space, to achieve the purpose of quality inspection. The experimental results show that this method can achieve convergence rate 18. 4 s, classification accuracy rate can reach 91. 6%, the overall performance is better than the traditional clustering methods, and has strong robustness.

Key words: mathematical morphology, granulometry, kernel function, clustering analysis

中图分类号: 

  • TP391