基于机器视觉的多相机视觉检测系统设计与实现

摘要在当今工业中，测量工件尺寸已成为一项重要的技术手段，它不仅可以帮助我们准确地评估工件的质量，还能够确保其合格。传统的人工测量尺寸方法存在诸多缺陷，如误差大、精度不高、耗时过长以及效率低下，这些都已经不能满足当今自动化技术的发展需求。随着科学技术的进步，机器视觉检测技术在提升效率、提升准确性以及增强智能化方面取得了显著成果，具备巨大的发展潜力。本文以前后端工件为研究对象，以机器视觉基本原理为基础，从工件尺寸检测的精度入手，对检测系统的通信选择及软件总体方案设计，界面设计，图像处理，相机标定和测量算法进行了详细分析，实现了工件尺寸的精确测量。主要研究内容有：首先，完成了机器视觉工件尺寸检测系统总体设计及硬件平台的构成选择，还有系统的软件设计，完成了基于MFC的人机交互界面的设计，按照图像采集系统的设计要求选择不同的通信并完成通信。然后，以图像处理为基本手段，采用Sobel算子进行图像滤波处理和选用Threshold算子进行图像阈值分割处理，并进行了开运算处理，获取到了高质量图像。完成了黑白棋盘法的相机标定，根据像素坐标转换为实际坐标和利用测量及计算公式得到工件尺寸。本文还使用了激光跟踪仪测量工件尺寸结果和机器视觉检测结果进行了对比和分析。最后，为了实现高精度的机器视觉检测的工件尺寸，提高工业生产线的自动化，拓宽机器视觉的应用，设计使用Visual Studio2017平台基于MFC框架下实现上位机（工控机)与机器人控制器的同步通信系统界面。该系统能够提高检测的精度和效率，适用于工业生产，具有一定的实用价值。关键词：机器视觉，I/O通信，网口通信，串口通信，相机标定，Sobel算子，Threshold算子，开运算，激光跟踪仪AbstractIn today's industry,the measurement of workpiece size has become an importanttechnical means,it can not only help us accurately assess the quality of the workpiece,butalso can ensure its conformity.The traditional manual measurement method has many defects,such as large error,low precision,too long time and low efficiency,which can not meet thedevelopment needs of automation technology.With the progress of science and technology,machine vision detection technology has made remarkable achievements in improvingefficiency,improving accuracy and enhancing intelligence,and has great potential fordevelopment.In this paper,the former back-end workpiece as the research object,based on the basicprinciple of machine vision,starting from the accuracy of the workpiece size detection,thedetection system communication selection and software overall scheme design,interfacedesign,image processing,camera calibration and measurement algorithm were analyzed indetail,to achieve the accurate measurement of the workpiece size.The main research contentsare as follows:First of all,complete the overall design of machine vision workpiece size detectionsystem and the composition of the hardware platform selection,as well as the software designof the system,complete the design of the human-computer interaction interface based onMFC,according to the design requirements of the image acquisition system select differentcommunication and complete the communication.Then,with image processing as the basic means,Sobel operator is used for imagefiltering processing and Threshold operator is used for image threshold segmentationprocessing,and open operation processing is carried out to obtain high quality image.Thecamera calibration with black and white checkerboard method was completed.The pixelcoordinates were converted into actual coordinates and the workpiece size was obtained byusing measurement and calculation formulas.In this paper,the laser tracker is used tomeasure the workpiece size and the machine vision test results are compared and analyzed.Finally,in order to realize the workpiece size of high-precision machine vision detection,improve the automation of industrial production line and broaden the application of machinevision,the Visual Studio2017 platfor