基于单目视觉的物体形状三维重建方法研究

发布时间：2018-04-27 19:40

本文选题：三维重建 + 单目视觉　；参考：《浙江大学》2017年硕士论文

【摘要】：三维重建是指对物体建立适合计算机处理的数字模型,是数字化表达客观世界的关键技术。随着3D打印、虚拟现实等技术的飞速发展,如何快速、简单地对物体进行三维重建不仅具有较高的科研价值,也蕴含着巨大的经济效益。本课题提出了一种基于单目视觉的低成本三维重建系统,该三维重建系统仅需要一个网络摄像头、一支红色线型激光笔和一个三正交背景板。操作者手持激光笔均匀地扫描放置在三正交背景板上的物体,并用摄像头拍摄整个扫描过程,经过图像处理提取激光条纹位置,并根据激光平面和投影射线的空间约束关系,计算物体表面的空间三维坐标。对点云进行滤波、重建、可视化等操作可建立物体的空间三维模型。本课题的主要研究成果如下:1、提出了一种基于单目视觉和正交平面几何约束关系的三维重建方法,该方法可以对非纹理物体进行精确重建。2、使用马尔科夫随机场模型对扫描得到的多层点云数据进行融合,该方法可以改善由于扫描不均匀造成的重建表面不平滑的现象。3、使用粒子群算法寻找最优的适合该三维重建系统的摄像头内部参数,改善了没有标定摄像头引起的误差,提高了重建结果的精度。
[Abstract]:Three-dimensional reconstruction refers to the establishment of a digital model suitable for computer processing of objects, which is the key technology of digital representation of the objective world. With the rapid development of 3D printing, virtual reality and other technologies, how to reconstruct objects quickly and simply has not only high scientific research value, but also great economic benefits. A low cost 3D reconstruction system based on monocular vision is proposed in this paper. The 3D reconstruction system only needs a webcam, a red linear laser pen and a triorthogonal background board. The operator, holding a laser pen, uniformly scans the object placed on the three-orthogonal background board, and takes the whole scanning process by using a camera. The position of the laser stripe is extracted by image processing, and according to the spatial constraint relationship between the laser plane and the projected ray, the position of the laser stripe is extracted by image processing. The three dimensional coordinates of the surface of the object are calculated. Three dimensional models of objects can be built by filtering, reconstructing and visualizing point clouds. The main research results of this thesis are as follows: 1. A 3D reconstruction method based on monocular vision and orthogonal plane geometric constraints is proposed. This method can reconstruct the non-texture object accurately. 2. The Markov random field model is used to fuse the scanned multi-layer point cloud data. This method can improve the non-smooth reconstruction surface caused by non-uniform scanning. The particle swarm optimization algorithm is used to find the best camera parameters suitable for the 3D reconstruction system, and the error caused by the uncalibrated camera is improved. The accuracy of the reconstruction results is improved.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【相似文献】