基于点云的数据处理技术及三维重建研究

发布时间：2018-12-13 10:53

【摘要】：随着空间数据采集设备的不断更新,单一的数据采集方式已不能满足当前三维重建的需要,如无人机航拍影像尺度大、视角广、现势性高,但缺乏建筑物立面信息;三维激光扫描仪采集到的点云数据精度高、地物信息丰富、无死角但大区域地形以及建筑物顶面点云采集与点云配准却十分困难。基于此,本文提出将无人机技术和三维激光扫描技术相结合,对两者的数据处理技术和重建方法进行了融合探讨。首先在对低空摄影测量系统总结的基础上,分析了无人机外业数据采集的流程及像控点的布设方法,并针对航拍影像获取点云数据的过程,对内业数据处理方法进行了探讨。其次,结合三维激光扫描仪的工作原理和外业数据采集方法,对标靶的摆设原则进行了分析,重点对多视点云的配准方法进行了论述,并通过设计不同实验方案,对比分析了不同配准方法的精度以及配种过程中不同分组数量对配准精度的影响。再次,结合现有的研究成果,重点探讨了点云数据的处理技术,分别从点云降噪、点云精简和点云分类三个方面进行分析研究。结合研究区域特征,探索了适用于研究区域的地物降噪方法和地形精简方法。最后,基于目前混合地形与地物三维重建模型精度低的不足,论文采用决策树的分类方法将三维重建过程划分为地形三维重建和地物三维重建两个可同时进行的过程,然后对两者进行组合,优化了重建方法,在保证建模精度的基础上提高了建模效率。
[Abstract]:With the continuous updating of spatial data acquisition equipment, the single data acquisition method can not meet the needs of 3D reconstruction. For example, UAV aerial photography image scale, wide angle of view, high present situation, but the lack of building facade information; The data of point cloud collected by 3D laser scanner have high accuracy, abundant information of ground objects, no dead angle but large area topography, and the collection of point cloud on top of building is very difficult. Based on this, a combination of UAV technology and 3D laser scanning technology is proposed, and the data processing technology and reconstruction method of the two technologies are discussed. Based on the summary of low altitude photogrammetry system, this paper analyzes the flow chart of UAV field data acquisition and the method of setting up image control points, and discusses the processing method of internal data in view of the process of aerial image acquiring point cloud data. Secondly, combining with the working principle of 3D laser scanner and the method of data acquisition in field of work, the principle of setting target is analyzed, and the registration method of multi-view cloud is discussed, and different experimental schemes are designed. The accuracy of different registration methods and the effect of different number of groups on registration accuracy are compared and analyzed. Thirdly, combined with the existing research results, this paper focuses on the point cloud data processing technology, respectively from the point cloud noise reduction, point cloud reduction and point cloud classification three aspects of analysis and research. Based on the characteristics of the study area, the methods of noise reduction and terrain reduction are explored. Finally, based on the lack of accuracy of 3D reconstruction model of mixed terrain and ground objects, the paper uses decision tree classification method to divide 3D reconstruction process into two simultaneous processes: terrain 3D reconstruction and ground object 3D reconstruction. Then the reconstruction method is optimized and the modeling efficiency is improved on the basis of ensuring the modeling accuracy.
【学位授予单位】：河北工程大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P225.2;P23

【参考文献】