室内近距离三维激光扫描技术研究

发布时间：2018-04-05 15:30

本文选题：三维激光扫描　切入点：激光条纹提取　出处：《江苏科技大学》2017年硕士论文

【摘要】：三维激光扫描技术一直是机器视觉领域的研究热点,通过对被测对象表面信息的采集,可以快速获取其表面的三维数据,完成三维场景的重建。本文研究了基于单摄像头和线激光的三维激光扫描技术,包括硬件平台设计、线激光条纹提取、参数标定和三维模型构建以及点云数据处理四个主要部分。本文首先采用线激光器对被测物体进行三维扫描,同时摄像头捕获含线激光的视频流数据,再通过线激光条纹亚像素定位及三角测距原理获取物体表面的点云数据,最后,对点云数据进行滤波、匹配和三维场景重建。论文的主要研究工作包括以下四个方面:1.完成三维激光扫描硬件平台的设计。在分析单点测距原理的基础上,对线激光扫描硬件平台进行了理论探讨,并据此对所用到的主要器件进行选型,包括摄像头、线激光器、舵机等,搭建了三维激光扫描平台。2.实现线激光条纹的识别和提取。根据线激光条纹较场景中其它物体具有更为显著的颜色特征,提出以图像中的红色通道分量作为研究对象,通过自适应阈值分割算法实现线激光条纹的粗提取,再用改进型加权重心法对线激光条纹进行精确定位,保证提取精度的同时,提高了线激光条纹的提取速度。3.参数标定和三维模型构建。分析摄像机成像模型,利用平面标定法实现摄像头参数的标定。同时结合三角测距原理和三维激光扫描硬件平台,分析线激光的扫描采样特点,完成对被测物体距离的测量和场景中三维点云数据的获取。4.点云数据处理。研究点云滤波特点,利用统计去噪算法对原始点云数据进行滤波处理,采用ICP(Iterative Closest Point)点云匹配算法完成点云密度的增强,再通过贪婪三角化算法实现点云三维模型的重建。
[Abstract]:3D laser scanning technology has been a hot topic in the field of machine vision. By collecting the surface information of the object under test, the 3D data of the object can be obtained quickly and the 3D scene can be reconstructed.In this paper, 3D laser scanning technology based on single camera and line laser is studied, including hardware platform design, line laser stripe extraction, parameter calibration and 3D model construction and point cloud data processing.In this paper, the measured object is scanned by a linear laser, and the video stream data are captured by the camera, then the point cloud data on the surface of the object are obtained by the sub-pixel location of the line laser stripe and the triangulation ranging principle.The point cloud data are filtered, matched and reconstructed.The main research work of the thesis includes the following four aspects: 1.The design of three-dimensional laser scanning hardware platform is completed.On the basis of analyzing the principle of single point ranging, the hardware platform of line laser scanning is discussed theoretically. According to this, the main devices used are selected, including camera, line laser, steering gear and so on, and a 3D laser scanning platform .2. is built.The recognition and extraction of line laser stripes are realized.According to the fact that the line laser stripes have more obvious color characteristics than other objects in the scene, this paper presents an adaptive threshold segmentation algorithm to realize the rough extraction of line laser stripes by taking the red channel component in the image as the research object.Then the improved weighted barycenter method is used to locate the linear laser stripes accurately, which ensures the extraction accuracy and improves the extraction speed of the linear laser stripes.Parameter calibration and three-dimensional model construction.The camera imaging model is analyzed and the camera parameters are calibrated by plane calibration method.At the same time, combining the principle of triangulation ranging and the hardware platform of 3D laser scanning, the characteristics of scanning sampling of line laser are analyzed, and the distance measurement of measured object and the acquisition of 3D point cloud data in scene. 4.Point cloud data processing.This paper studies the characteristics of point cloud filtering, uses statistical denoising algorithm to filter the original point cloud data, uses ICP(Iterative Closest Point point matching algorithm to enhance the point cloud density, and then realizes the reconstruction of point cloud 3D model by greedy triangulation algorithm.
【学位授予单位】：江苏科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TN249

【相似文献】