摘要本文描述了一种基于机器视觉的圆形零件在线实时检测的软件及其实现。软件基于MV-VS2000机器视觉平台设计。由CCD摄像头采集流水线上零件的图像，通过阈值分割后进行二值化分离出感兴趣的零件部分的图像，然后使用差影法结合链表同时提取一张图像内所有零件的轮廓，并使用最小二乘法拟合出各个圆形零件外廓圆的表达式，最终得到零件的半径、面积、圆心位置等信息。此外该软件还可以利用差异阈值法实现光滑零件表面的划痕检测。经过MV-VS2000平台的实际测试，该软件算法实现了实时的非接触式检测。实验结果表明其在误差1%的前提下1分钟内至少可以检测100个工件，既可以测量尺寸信息也可以检测出残次品的存在。82353

毕业论文关键词：机器视觉  数字图像处理  轮廓提取  多圆检测  划痕检测

毕业设计说明书（论文）外文摘要

Title          Software design of workpiece on-line measurement based on machine vision                               

Abstract This thesis represents a circular workpiece measurement software system based on machine vision。 This software is designed based on MV-VS2000 machine vision platform。 Pictures of workpiece are captured by a CCD camera。 And the areas of workpiece which we are interested are separated through threshold pision and the following binarization。 Then we use background subtraction combined with linked lists to extract the contour of all workpiece in one picture simultaneously。 At last, by using least square method, we derive the equation of every circular workpiece's contour and then their radius, area, center point position and so on can be deduced from the equation。 In addition, the software can also detect scratch on smooth surface by difference threshold method。 After the real tests on MV-VS2000 platform, we successfully achieves real-time non-contact detection using this software。 The result indicated the software is able to measure at least 100 workpiece within 1min and with the measurement error under 1%。 It is qualified for workpiece measurement and can also be used for defective products detection。

Keywords  machine vision, digital image processing, contour extraction, multiple-circle detection, scratch detection 

目   次

1  绪论 1

1。1  机器视觉系统研究的意义 1

1。2  机器视觉系统的发展现状 1

1。3  论文主要研究内容及工作 2

2  基于机器视觉的零件在线检测系统方案 3

2。1  系统结构 3

2。2  系统的硬件配置 3

2。2。1  机械运动模块 4

2。2。2  视觉模块 4

2。2。3  工控组态模块 6

2。3  系统的软件配置 6

2。4 研究方法及流程 6

2。5 本章小结 7

3  系统数字图像处理的方法 8

3。1  零件图像的分割 8

3。1。1  中值滤波 9

3。1。2  阈值分割 9

3。2  零件轮廓的提取