三维激光测量仪的关键技术研究

发布时间：2018-03-07 07:56

本文选题：三维扫描测量　切入点：图像处理　出处：《吉林大学》2015年硕士论文　论文类型：学位论文

【摘要】：本文对于三维激光测量仪的关键技术进行了研究，包括系统结构、二维图像处理、三维坐标重建、点云后期处理等几个技术方面。针对每一项关键技术都说明了其技术路线，制定了解决方案，并且实现了每一部分的功能，完成了三维激光测量仪的整体设计与关键技术的研究工作。采用上下位机串口通信的方式建立了系统结构，设计了基于步进电机的扫描式测量方式，基于线激光光源采用了光刀扫描的测量方法，运用CCD相机获取结构光图像，从而完成了三维激光测量仪的结构设计。在二维图像处理中，采用经典的steger算法对于结构光图像的条纹中心进行了提取，又利用网格分割的方式减少了图像处理的计算量，简化了算法，并输出了二维坐标集合的结果形式。在论文中，对于一幅结构光照射于人体的图像进了二维图像处理，测量结果比较理想，验证了算法的可行性。在三维重建中，利用已知尺寸参数的黑白棋盘格定标板对系统的图像放大倍数进行了标定，根据旋转扫描测量的结构，结合空间立体几何相关知识，推导出二维像素点坐标的三维尺寸坐标重建算法。设计了三个传感器的测量结构，实现对被测物体侧面的360度测量，利用坐标系的旋转推导了三个传感器的三维重建算法，获取了点云的初始数据。在对点云进行后期处理的研究工作中，分析了系统模型和算法的误差来源，，根据旋转式扫描中物距为变参量这一问题，指出了三维重建算法存在的不足，提出了修改方案。采用了变物距的迭代修正算法，对点云初始数据进行迭代修正，并通过实验对算法进行验证。实验中，对原始点云数据抽样，利用抽样点计算目标函数，将计算的函数值与测量真实值进行对比。通过多次修正，验证了该算法可以减小测量误差，并且具有收敛性。对于多点云的拼接问题，在经典的ICP算法原理基础上，采用改进的ICP算法，计算“点对”的权重，提高了运算效率。采用M-估计改进了目标函数，从而去掉了空间异常点的干扰，使得点云数据更加精确，提高了三维测量仪的测量精度。最后对于三维测量仪的关键技术进行了总结与展望，提出了测量仪的改进空间和发展前景。本文从结构、各组件的参数、技术路线、算法原理与验证等多方面对三维激光测量仪的关键技术进行了研究，分析了各个关键技术的优点与不足，提出并实现了其解决方案，系统全面地研究论证了三维激光测量仪的实现方案，为三维激光测量仪的产业化生产研发提供了有效的技术依据。希望通过本文对于三维测量技术的研究工作，能让人们了解三维激光测量仪的构造，原理和功能，能给相关领域的科学工作者带来借鉴的意义，推动促进三维测量产业的发展。
[Abstract]:In this paper, the key technologies of 3D laser measuring instrument are studied, including system structure, 2D image processing, 3D coordinate reconstruction, point cloud post-processing and so on. The solution is made and the function of each part is realized. The whole design of 3D laser measuring instrument and the research of key technology are completed. The structure of the system is established by serial communication between upper and lower computers. The scanning measurement method based on stepping motor is designed. The scanning method of optical knife is used based on the line laser light source, and the structured light image is obtained by CCD camera. Thus, the structure design of the three-dimensional laser measuring instrument is completed. In two-dimensional image processing, the classic steger algorithm is used to extract the fringe center of structured light image, and the grid segmentation is used to reduce the computation of image processing and simplify the algorithm. The result form of two-dimensional coordinate set is outputted. In this paper, an image of human body irradiated by structured light is processed, and the measurement result is ideal, which verifies the feasibility of the algorithm. In 3D reconstruction, the image magnification of the system is calibrated by using the black-and-white checkerboard calibration board with known size parameters. According to the structure of the rotational scanning measurement and the related knowledge of spatial solid geometry, the image magnification of the system is calibrated. The 3D dimension coordinate reconstruction algorithm of two-dimensional pixel coordinate is derived. The measurement structure of three sensors is designed to measure 360 degrees of the measured object side, and the 3D reconstruction algorithm of the three sensors is deduced by the rotation of coordinate system. The initial data of the point cloud is obtained. In the research of point cloud post-processing, the error source of the system model and algorithm is analyzed. According to the problem that the distance is variable parameter in rotary scanning, the deficiency of 3D reconstruction algorithm is pointed out. A modified scheme is proposed. The point cloud initial data is modified iteratively by using the variable distance iterative correction algorithm, and the algorithm is verified by experiments. In the experiment, the object function is calculated by sampling the original point cloud data, using the sampling point to calculate the objective function, The calculated function value is compared with the measured real value. Through several revisions, it is verified that the algorithm can reduce the measurement error and has convergence. For the multi-point cloud splicing problem, on the basis of the classical ICP algorithm principle, By using the improved ICP algorithm, the weight of "point pair" is calculated, and the operation efficiency is improved. The objective function is improved by using M- estimation, which eliminates the interference of the spatial outliers and makes the point cloud data more accurate. The measuring accuracy of the three-dimensional measuring instrument is improved. Finally, the key technology of the 3D measuring instrument is summarized and prospected, and the improvement space and development prospect of the measuring instrument are put forward. In this paper, the structure, the parameters of each component and the technical route are discussed. In this paper, the key technology of 3D laser measuring instrument is studied from the aspects of algorithm principle and verification, the advantages and disadvantages of each key technology are analyzed, and the solution is put forward and realized. The realization scheme of 3D laser measuring instrument is studied and demonstrated systematically and comprehensively, which provides an effective technical basis for the industrial production and development of 3D laser measuring instrument. It is hoped that through the research work of 3D measurement technology in this paper, It can make people understand the structure, principle and function of 3D laser measuring instrument, can bring reference significance to scientists in related fields, and promote the development of 3D measurement industry.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.41;TN249

【参考文献】