一种基于机器视觉的轮胎胎面检测系统

发布时间：2018-03-13 11:17

本文选题：机器视觉　切入点：图像处理　出处：《杭州电子科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着工业自动化发展,现代化生产过程对检测方法、精度和速度提出了更高的要求。机器视觉检测系统及技术的广泛应用符合工业自动化发展的需要,能有效提高检测的效率和精度,从而提高工业的自动化程度。在橡胶轮胎的生产中,需要在线对胎面进行长度检测,现有的检测方法大都采用人工在线抽样检测并记录检测结果,其检测效率较低,检测精度也不能得到有效保障。本文针对橡胶轮胎胎面生产线的要求,结合机器视觉检测技术,设计了一种轮胎胎面检测系统。该系统可实现生产线上的橡胶轮胎胎面的几何量(长度和倾斜角度)、胎面字符等信息的在线检测与识别,并将检测结果实时保存到上位机数据库。本文主要完成了以下工作:(1)对轮胎胎面检测系统进行总体方案设计。将检测系统分为图像采集、信息处理、信息执行三个模块,并进行任务分配,对各模块中的关键技术及基本原理进行了阐述。(2)设计检测系统的几何量检测方法。基于机器视觉技术,采用图像预处理、阈值分割、胎面区域提取、边缘检测等方法实现了基于形状特征的长度检测方法并进行改进;设计了基于霍夫变换的倾斜角检测方法;从检测准确率、检测时间方面对方法的性能进行了分析。(3)设计检测系统的字符检测方法。基于机器视觉软件和机器学习方法,通过图像预处理、字符训练、字符分类器、字符识别等步骤实现字符检测;从检测准确率、检测时间方面,分析了不同分类器对字符检测的影响。(4)根据检测系统的要求,结合下位机的通讯协议、MySQL数据库,在VS2010和Halcon平台上进行编程,实现检测系统的完整功能。在检测现场对该系统的准确率、实时性和检测精度进行测试。
[Abstract]:With the development of industrial automation, modern production process has put forward higher requirements for testing methods, accuracy and speed. The wide application of machine vision inspection system and technology meets the needs of the development of industrial automation. Can effectively improve the efficiency and accuracy of the detection, thereby increasing the degree of industrial automation. In the production of rubber tyres, it is necessary to check the tread length online. Most of the existing detection methods use manual on-line sampling and record the test results, its detection efficiency is low, and the detection accuracy can not be effectively guaranteed. In this paper, according to the requirements of rubber tire tread production line, combined with machine vision detection technology, A tire tread detection system is designed, which can be used to detect and recognize the geometry of rubber tire tread (length and tilt angle, tread character, etc.) on line. The system is divided into three modules: image collection, information processing and information execution, and the system is divided into three modules: image collection, information processing and information execution. The key technology and basic principle of each module are described in detail. Based on machine vision technology, image preprocessing, threshold segmentation and tread area extraction are adopted. Edge detection and other methods have realized and improved the length detection method based on shape features, and designed a slope angle detection method based on Hough transform. Based on machine vision software and machine learning method, image preprocessing, character training, character classifier, Character recognition and other steps to achieve character detection, from the detection accuracy, detection time, analysis of different classifiers on character detection. 4) according to the requirements of the detection system, combined with the communication protocol of the lower computer MySQL database, The complete function of the detection system is realized by programming on the platform of VS2010 and Halcon, and the accuracy, real time and accuracy of the system are tested on the spot.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】