载料带高速视觉检测系统设计与关键技术研究

发布时间：2018-05-03 09:53

本文选题：载料带 + HALCON　；参考：《湖北工业大学》2017年硕士论文

【摘要】：载料带是微电子行业中的一种使用率特别频繁的包装卷材,有越来越多的载料带生产企业在制造过程里了不同的自动检测装置,以确保产品质量的可靠性很高。以前的检测方式通常是人工检测,但人工检测存在检测速度慢,检测精度低等问题,不能保证制造的可靠性,也不能满足现代制造的要求。本文以载料带的一些常见缺陷为检测对象,对载料带在线检测的高速视觉检测系统以及载料带缺陷快速检测和识别的关键技术做了较为系统的研究,希望能用机器视觉技术代替人工检测,主要研究内容如下:(1)检测系统的设计。针对在线检测的需求,确定了检测系统的结构,工作流程和检测方案;结合载料带缺陷检测精度的要求,确定出系统硬件所需的光源、相机与镜头及图像采集卡;确定了系统软件平台HALCON。(2)图像特征区域的快速定位算法。在获得载料带高质量图像的基础上,对图像进行预处理,这一过程包括图像的灰度化,中值滤波,阈值分割,然后介绍了图像拼接的主要步骤和基本原理、基于Harris算法的主要流程、一种改进的图像拼接方法;介绍了两种图像特征区域快速定位的计算方法:基于图像灰度值累计分布的计算和分析,以及计算图像一维矩的快速算法;给出了基于Harris算法和本文采取的算法之间的对比结果,通过结果对比表明本文算法在拼接速度和匹配速度上都有所提高。(3)载料带缺陷的快速检测和识别。总体上分为基于形状特征的模板匹配法和基于灰阶分析的检测法。基于形状特征的模板匹配法是基于HALCON对目标图像进行模板匹配,然后再进行图像差分;阐述了基于HALCON的三种模板匹配技术,图像差分法的基本原理及其算法的主要流程;基于灰阶解析的检测方法,包括图像的灰度化处理、图像的灰度直方图分析以及Blob分析法。
[Abstract]:Load belt is a kind of packaging coil with high utilization rate in the microelectronics industry. More and more enterprises have different automatic detection devices in the manufacturing process to ensure the high reliability of product quality. The former detection method is usually manual detection, but manual detection has some problems such as slow detection speed and low detection precision, which can not guarantee the reliability of manufacturing, nor can it meet the requirements of modern manufacturing. This paper takes some common defects of load belt as the detection object, and makes a systematic research on the high speed visual inspection system of load belt on-line detection and the key technology of fast detection and recognition of load belt defect. It is hoped that machine vision technology can be used instead of manual detection. The main research contents are as follows: 1) the design of detection system. According to the demand of on-line detection, the structure, work flow and detection scheme of the detection system are determined, and the light source, camera, lens and image acquisition card for the hardware of the system are determined according to the requirements of the inspection accuracy of the material carrying belt defect detection. The fast localization algorithm of HALCON.-2) image feature region is determined. On the basis of obtaining the high quality image of the load belt, the image is preprocessed. This process includes grayscale, median filter, threshold segmentation, and then introduces the main steps and basic principles of image stitching. Based on the main flow of Harris algorithm, an improved image stitching method is introduced, and two computing methods of fast location of image feature region are introduced: the calculation and analysis based on the cumulative distribution of image gray value, and the fast algorithm for calculating one-dimensional moment of image. The comparison results between the Harris algorithm and the algorithm adopted in this paper are given. The results show that the proposed algorithm can improve the speed of stitching and matching. In general, there are template matching method based on shape feature and detection method based on gray scale analysis. The template matching method based on shape feature is to match the target image with template based on HALCON, then the image difference is carried out. Three template matching techniques based on HALCON, the basic principle of the image difference method and the main flow chart of the algorithm are expounded. The detection method based on gray scale analysis includes image grayscale processing, image grayscale histogram analysis and Blob analysis.
【学位授予单位】：湖北工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN40;TP391.41

【参考文献】