【作者】 王凡；

【导师】 黄勋；

【作者基本信息】 陕西科技大学 ， 机械工程， 2012， 硕士

【摘要】 基于机器视觉技术，本课题对现有的皮革喷涂机进行了改造，设计了一种集图像采集、图像处理和自动控制喷涂的智能控制系统。本课题设计的智能控制系统——基于机器视觉的皮革喷涂控制系统主要由以下硬件构成：计算机、光源、工业面阵相机、数字图像采集卡和数字量输入输出卡（DI/DO卡）。在喷涂机的传输系统前端安装相机，以采集待喷涂皮料的图像信息；相机拍摄数字图像经由数字图像采集卡传输到计算机中，按采集的时间顺序对其进行分析处理；图像处理完成后，计算机根据图像处理的结果生成控制信号，并将此信号传输到DI/DO卡中，由DI/DO卡控制喷枪状态。基于机器视觉的皮革喷涂控系统的系统软件采用微软公司VC++6.0进行编写，其主要由五部分组成：图像采集模块、图像预处理模块、区域快速识别模块、喷涂数据生成模块和喷涂控制模块。为满足皮革喷涂系统图像处理部分的实时性，本设计在系统主线程序的基础上，采用多线程的方式，实现图像采集、图像处理和显示以及PCI2320的中断。出于对喷涂生产过程以及喷涂环境的考量，本课题中图像处理过程的算法确定，利用了模拟实验，对比了常用的图像处理算法，选取了适合本设计的方法。最终确定的图像处理方案为：采集到的图像先用高斯去噪法去噪，然后利用经验法灰度阈值方法进行图像分割，最后图像后处理的部分选用了灰度闭运算的方法。本设计中使用了高检测精度的机器视觉技术，实现了皮革形状的精确检测。并且，根据形状检测的结果，利用DI/DO卡快速便捷地控制各个喷枪，实现了图像采集和涂料喷涂实时同步运行。更多还原

【Abstract】 Based on machine vision technology, leather spraying machine wasreformed in our work. And a kind of intelligent control system whichincludes image acquisition, image processing and automatic controlspraying gun, was designed.The leather spraying control system based on machinery visionmainly includes computer, camera, image acquisition card, light source,DI/DO card of digital signal and leather transmission institutions. Theproposed execution process was developed as follows: we installed acamera before transmission system collected the image information ofleather. The image collected through image gathering card wastransmitted to a computer and then be analyzed. After the analysis, thetreatment was produced automatically by computer and transmitted toDI/DO card. The condition of the spraying gun was controlled byDI/DO card.The interface was designed and the function was realized by usingMicrosoft VC++6.0in leather spraying measurement system. Thesoft-wares of leather spraying system based on Machine vision wereimage acquisition module, image processing module, regional rapididentification module, spraying data generating module and sprayingcontrol module. To meet with system image processing requirements forreal-time, a multi-threading was used at the basis of system programwhen designing soft-wares. Multi-threading included GrabThread imagecollection line, image processing and display thread and PCI2320threadinterrupt.The common image processing algorithms was compared in thesimulation experiment. And the design method was finally determinedbased on the reality of this subject. The final design method was that: The collected image was firstly denoised using Gauss de-noising, andthen divided using empirical threshold for image segmentation. At last,the image was dealt with gray closing operation method for postprocessing part.Detection precision machine vision system was used to realize theaccurate detection of leather shape in our work. And, each gun wascontrolled fast and conveniently by using DI/DO card according to theshape detection results, to achieve the synchronous operation of imageacquisition and paint spraying real-time.更多还原