零件几何尺寸精密测量系统设计与实现

发布时间：2018-02-12 09:26

本文关键词： 图像处理亚像素高精度零件尺寸测量批量测量　出处：《西南科技大学》2016年硕士论文　论文类型：学位论文

【摘要】：随着生产与科技的发展,加工精度的提高和加工对象的变化,人们对测量精度、效率以及自动化程度要求越来越高。实现零件形状尺寸的快速、精确测量是零件几何尺寸测量系统中最为关键的问题。本文针对传统测量在小型、柔软、形状特异等零件测量上的不足,运用亚像素边缘检测、图像配准、相机标定等关键的计算机视觉技术,开发通用零件尺寸高精度在线自动化测量系统。主要工作如下:1、亚像素边缘检测算法研究。着重研究了基于亚像素检测的边缘定位技术,设计实验对比拟合法、灰度矩和空间矩等亚像素检测算法的精度和效率,定量分析影响精度的主要因素,选择能满足实时测量中精度、速度、稳定性兼顾的亚像素检测和边缘定位算法。2、高精度零件图像配准算法设计。设计了一种基于质心和最小外接矩形的快速高精度零件图像配准算法。分为初始配准和精确配准,初始配准通过利用零件最小外接矩形和图像形心,缩减配准参数范围提高配准速度。精确配准将互信息作为相似度准则提高配准精度。实验结果表明本算法能同时在时间效率和匹配精度上达到批量测量的要求。3、零件几何尺寸精密测量系统开发。分析了图像测量系统的软硬件结构,应用基本的亚像素边缘检测、图像配准等技术,设计开发出零件几何尺寸测量系统,具有单品测量、批量检测、统计分析等功能。测试结果表明,系统重复测量精度能够达到5μm。
[Abstract]:With the development of production and science and technology, the improvement of machining precision and the change of machining object, people demand more and more high measuring precision, efficiency and automation degree. Accurate measurement is the most important problem in the geometric dimension measurement system of parts. In this paper, the sub-pixel edge detection and image registration are used to solve the problems of traditional measurement in small, soft and shape specific parts. The key computer vision technology, such as camera calibration, is used to develop a high precision on-line automatic measuring system for the dimensions of common parts. The main work is as follows: 1, the research of sub-pixel edge detection algorithm, and the research of edge location technology based on sub-pixel detection. The precision and efficiency of sub-pixel detection algorithm, such as analogy method, gray moment and spatial moment, and quantitative analysis of the main factors affecting accuracy are designed, and the accuracy and speed of real-time measurement are selected. A stable sub-pixel detection and edge location algorithm .2. high precision part image registration algorithm is designed. A fast high-precision part image registration algorithm based on centroid and minimum external rectangle is designed, which is divided into initial registration and accurate registration. The initial registration is made by using the minimum external rectangle of the part and the center of the image, The precision of registration is improved by reducing the range of registration parameters and improving the registration speed by using the accurate matching information as the similarity criterion. The experimental results show that the algorithm can meet the requirements of batch measurement in both time efficiency and matching accuracy. The development of precise measuring system for geometric dimensions of parts. The hardware and software structure of the image measurement system is analyzed. Based on the basic techniques of sub-pixel edge detection and image registration, the geometric dimension measurement system of parts is designed and developed, which has the functions of single product measurement, batch detection, statistical analysis and so on. The test results show that the precision of repeated measurement of the system can reach 5 渭 m.
【学位授予单位】：西南科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.41

【相似文献】