基于激光扫描和SFM的非同步点云三维重构方法

室外场景具有测量数据量大、扫描数据易重叠及建筑物表面信息复杂等特点,单靠激光扫描方法能够获得场景精确的深度信息,但缺乏颜色和纹理信息,利用从运动中恢复结构(SFM)方法可获得丰富的彩色信息,但重构精度不高,若将两种设备固定进行在线实时同步测量,易受到测量环境和系统制约不易实现。针对此问题,提出了一种基于激光扫描和SFM结合的非同步点云数据融合的三维重构方法。首先,提出利用手动选择控制点进行7自由度初始配准,再利用迭代最近点(ICP)算法对初始配准结果进行精确配准,最后利用最近点搜索算法将分布在经基于面片的多视图立体视觉(PMVS)算法优化后的SFM数据中的颜色信息与激光扫描的点云坐标进行融合。实验结果和数据分析显示,本文的方法能有效地将激光扫描与SFM点云数据进行融合,实现了室外大场景的三维彩色重构。

Non synchronous point cloud algorithm for 3D reconstruction based on laser scanning and SFM

The outdoor scene has the characteristics of large measurement data, easy scanning data overlapping and complex building surface information. Laser scanning method can obtain accurate scene depth information, however its weakness is lacking of object surface texture and color information. The camera based SFM method can obtain rich color information, however its reconstruction accuracy is poor. It is a probable solution to fix the two kinds of equipment and conduct online real time synchronous measurement, due to the restriction of measurement environment and system this idea can not be realized easily. Aiming at above mentioned problem, this paper proposes a non synchronous point cloud fusion algorithm for 3D reconstruction based on 3D laser scanner and camera based SFM. Firstly, the method of manually selecting control points is used to conduct the initial registration with 7 degrees of freedom, and the ICP algorithm is used to perform the high precise registration based on the initial registration result. At last, the nearest point search algorithm is used to fuse the color information distributed in the SFM data that are optimized with PMVS algorithm and the point cloud coordinate information obtained from laser scanning to achieve the 3D reconstruction of large outdoor scene. The 3D reconstruction experiments and data analysis show that the proposed algorithm can effectively fuse the two kinds of point cloud data come from laser scanner and camera based SFM to achieve the 3D color reconstruction of outdoor large scene.