光子学报, 2018, 47 (11): 1112002, 网络出版: 2018-12-17
基于单目视觉的曲面弧度测量方法
Arc Measurement Method of Curved Surface Based on Monocular Vision System
光学技术与仪器 曲面弧度测量 单目视觉 视觉测量 三维重建 Optical technology and instrument Curvature measurement of curved surface Monocular vision Visual measurement 3D reconstruction
摘要
针对异形件曲面弧度的测量问题, 提出了一种基于单目视觉的曲面弧度测量方法.首先, 用基于三色LED环形光源和CCD彩色相机组成的单目视觉系统采集单张包含被测曲面表面特性的彩色图像; 其次, 分析该图像的彩色分布特征、灰度变化规律和相邻区域特征的相关性, 设计分割方法, 完成目标曲面的分割; 然后, 基于单目视觉系统成像模型, 结合材质与实验环境相关参量, 推导出被测表面高度计算方程; 最后, 解出相邻像素点间的高度累加分量, 进而还原被测点的高度, 在此基础上, 建立曲面弧度测量模型并完成曲面弧度测量.分别对3D打印曲面样品、螺钉螺纹、SIM卡槽进行了曲面弧度测量实验, 与激光三角法测量结果对比, 所提方法误差小于±3.6%, 测量准确度为0.000 1 rad, 曲面弧度测量时长小于0.6 s.实验结果表明, 该方法对测量物体边缘、高度突变, 高度渐变等类型的曲面都具有良好的适应性, 在实现较高准确度测量的同时, 具有较快的测量速度.
Abstract
A curved surface measurement method based on monocular vision system is proposed to measure the curved surface of the special-shaped parts. First, the monochromatic vision system which is composed of a tricolor LED ring structure light source and a three-color CCD color camera is used to collect a single color image that contains the surface characteristics of the measured surface. Secondly, the correlation of color distribution features of the images, the regular of gray level variation and characteristics of adjacent regions are analyzed, and the segmentation method is designed to complete the target surface segmentation. Then, the calculation equation of the measured surface height is derived based on the image model of the monocular vision system and the related parameters of the material and the experimental environment. Finally, the height of the measured points is reduced by solving the height accumulation component between adjacent pixels. On this basis, the surface curvature measurement model is established and the curved surface is measured. The surface curvature measurement experiments respectively for 3D printing surface samples, screw thread and SIM card slot are carried out based on the proposed method. Comparing with the laser triangulation measurement method results, it shows that the errors are all less than 3.6%, the measurement accuracy is 0.000 1 rad, and the time length of measuring is less than 0.6 s. The experimental results show that the proposed method has good adaptability for measuring object edge, height abrupt change, height gradual change and other types of surface. It can achieve high precision measurement and fast measurement speed.
吴福培, 郭家华, 李昇平, 叶玮琳, 张宪民. 基于单目视觉的曲面弧度测量方法[J]. 光子学报, 2018, 47(11): 1112002. WU Fu-pei, GUO Jia-hua, LI Sheng-ping, YE Wei-lin, ZHANG Xian-min. Arc Measurement Method of Curved Surface Based on Monocular Vision System[J]. ACTA PHOTONICA SINICA, 2018, 47(11): 1112002.