基于传统图像算法与软件工程的机器视觉平台应用研究

On Machine Vision Platform Based on Traditional Image Algorithm and Software Engineering

为了解决深度学习在尺寸测量、视觉引导等领域中呈现出较低的检测准确度的问题,基于经典图像算法,将其模块化定制为具体工具,以供第三方软件调用,从而设计了一种精准测量机制。首先,设计并开发多种视觉工具,包括灰度统计、直线卡尺、颜色识别、直边检测、弧边检测、条码识别、字符OCR、斑块分析、边缘提取、找点工具等,并将这些工具集成在机器视觉平台中;然后,结合用户使用习惯,规划平台软件系统的功能模块,包括图像显示区、结果显示区、流程工具区、机器视觉工具区和通信工具区,并基于拖拽式交互方式,实现用户无需编程即可完成项目开发与应用的目的;最后,以视觉引导应用,基于机器视觉平台,完成图像采集、定位、测量、数据保存等应用。最后,通过实验测试发现:本文开发的平台具有更高的检测精度,可准确定位出缺陷区域。

In order to solve the problem of low detection accuracy in the fields of dimension measurement and visual guidance,this study is based on the independent research and development of traditional image algorithm,modular customization as a specific tool for third-party software call,forming an accurate measurement mechanism.Firstly,we design and develop the vision tools,including gray statistics,linear caliper,color recognition,straight edge detection,arc edge detection,bar code recognition,character OCR,patch analy⁃sis,edge extraction,point finding tools,and integrate the tools into the machine vision platform.Then,combined with users'habits,the functional modules of the platform software system are planned:image display area,result display area,process tool area,machine vi⁃sion tool area and communication tool area.Based on the drag and drop interaction mode,users can complete the project development application without programming.Finally,based on the machine vision platform,the application of vision guidance is used to complete image acquisition,positioning,measurement and data storage.The experimental test results show that the platform developed in this pa⁃per is conducive to the landing of scenarios with high decision-making requirements,and can ensure the rapidity and effectiveness of the project.