数控V割机机器视觉定位方法研究

发布时间：2019-02-21 13:03

【摘要】：本文以某公司生产的数控V割机作为参考对象。该V割机的载物平台只具有一个自由度,待切割PCB只能随着载物平台在一个方向上做直线往复运动。在正式切割之前,通过人工测量出PCB的定位孔心到其试刀线的距离作为V割机数控系统的修正值,并将这个值通过计算机交互界面输入V割机的系统控制程序中。通过这种方式将数控V割机的系统原点与机械原点重合,从而达到定位的目的。因此该数控V割机原有的定位方法主要是由人工对PCB的一些特征尺寸的测量来实现的。显然,由于人工的测量引入了较大误差,致使V割机很难达到最终的定位精度要求,并且降低了工厂的生产效率。本文针对该数控V割机的工作原理以及原有定位方法的缺点,设计了一个机器视觉定位系统来代替原有的人工定位操作,并提出两套测量方案。它包括硬件系统与软件系统。硬件系统由摄像机、光源、电移台、光栅尺以及载物平台组成;软件系统包括图像采集、图像处理、摄像机标定、光栅尺读数等模块。为保证测量精度与较小的工作距离,本文所设计的定位测量系统分两次采集PCB图像,两次采集之间通过电移台的移动实现对PCB的两处定位标记分别拍摄。第一个方案是借助光栅尺测量电移台的位移和两次采集到的图像数据,间接计算出PCB的定位孔心到其试刀线的距离;第二个方案是通过图像拼接技术把初、末位置采集的两幅PCB图像拼接成为一幅镶嵌图像,然后在镶嵌图像中直接计算出该距离。文本主要包括以下几个方面的内容:一、针对数控V割机的定位原理以及定位精度的要求,设计了一个机器视觉定位系统,并对其硬件系统与软件系统进行了具体的设计。二、根据视觉定位的需要,必须首先对摄像机进行标定,目的是消除图像畸变以校正图像。本文以摄像机模型为依据,详细描述了摄像机标定的原理、过程以及标定结果。接着通过基于双目标定的图像拼接技术将初、末位置的PCB图像拼接成一幅镶嵌图像。最后通过对标准量块的测量验证了摄像机标定的结果。三、本文在对提取PCB的定位标记时用到的一些图像处理算法进行了详细地分析并通过程序实现了这些算法。最后在校正后的PCB图像中对定位标记进行测量。四、编写人机交互界面。
[Abstract]:This paper takes the NC V cutting machine produced by a company as the reference object. The loading platform of the V cutting machine has only one degree of freedom, and the PCB to be cut can only be reciprocated in a straight line with the loading platform in one direction. Before the formal cutting, the distance between the center of the PCB location hole and its test cutter line is measured manually as the revised value of the NC system of the V-cutting machine, and the value is input into the system control program of the V-cutting machine through the interactive interface of the computer. In this way, the system origin of the NC V cutting machine is coincident with the mechanical origin, so as to achieve the purpose of positioning. Therefore, the original positioning method of the NC V cutting machine is mainly realized by manual measurement of some characteristic dimensions of PCB. It is obvious that due to the large error introduced by manual measurement, it is difficult for the V cutter to meet the final positioning accuracy requirement and reduce the production efficiency of the factory. In this paper, a machine vision positioning system is designed to replace the original manual positioning operation in view of the working principle of the NC V cutting machine and the shortcomings of the original positioning method, and two sets of measurement schemes are proposed. It includes hardware system and software system. The hardware system consists of a camera, a light source, an electric moving platform, a grating ruler and a carrier platform, and the software system includes image acquisition, image processing, camera calibration, grating ruler reading and so on. In order to ensure the accuracy of the measurement and the small working distance, the positioning and measuring system designed in this paper collects the PCB images twice, and the two positioning markers of the PCB are photographed separately by the moving of the electric moving platform between the two times. The first scheme is to measure the displacement of the electric moving table and the image data collected twice by using the grating ruler to calculate indirectly the distance between the locating hole center of PCB and its test tool line. In the second scheme, two PCB images collected at the initial and last positions are stitched into a mosaic image by image mosaic technology, and the distance is calculated directly in the mosaic image. The text mainly includes the following aspects: firstly, a machine vision positioning system is designed according to the positioning principle and positioning accuracy of the NC V cutting machine, and the hardware and software systems are designed in detail. Secondly, according to the need of visual positioning, the camera must be calibrated first, in order to eliminate the image distortion to correct the image. Based on the camera model, the principle, process and results of camera calibration are described in detail in this paper. Then the PCB image at the beginning and end position is stitched into a mosaic image by using the image mosaic technology based on double target determination. Finally, the result of camera calibration is verified by measuring the standard block. Thirdly, some image processing algorithms used in the extraction of PCB localization markers are analyzed in detail, and these algorithms are realized by program. Finally, the location mark is measured in the corrected PCB image. Fourth, compile the man-machine interface.
【学位授予单位】：天津科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TG659

【参考文献】