地面三维激光扫描点云分辨率研究

Research on the Point Cloud Resolutions of TLS

分辨率决定了从点云中提取目标细节的能力,可分为平面分辨率、距离分辨率和强度分辨率。分析了在分辨率中起主导作用的平面分辨率的主要影响因素,得出了计算平面分辨率的理论公式并通过实验验证了其正确性,对地面三维激光扫描实际应用具有指导意义。

TLS(Terrestrial Laser Scanning)uses laser to measure the 3D coordinates of a given region of an objeet's surface automatically,in a systematic order at a high rate in near real time.TLS is a method which can get the 3D model of the object directly.Due to its high speed in obtaining unstructured point clouds from all surfaces within the field of view,TLS can be used in many fields such as deformation monitoring,3D modeling of building or other large scale scene. The data from TLS have the format of point cloud.Compared with traditional survey method,TLS focuses not only on the accuracy of single point position but also on the modeling accuracy of the 3D surface.Therefore,some resolutions are im- portant characters in the point cloud.The resolutions govern the level of identifiable details within the scanned point cloud.These resolutions can be classified into angular,range and intensity resolution.The first one is like an indicator of resolutions and is the focus of this study.The main issues concern the impact of angular resolution by different scan range and angle.Based on the principle of TLS,the basic relations and formulas concerning the accuracy are put forward for dif- ferent scan range and angle.And then,a Leica HDS3000 laser scanner is used to testify in the field. The first test is concerned with the change of accuracy by different scan range.A piece of plane ground is selected for scanning,and a rectangle range is cut from the point cloud to calculate the histogram of distances from the scan points to the scanner.Contrast with the theoretical number of scan points,the curves calculated appear quite similar.The result can be explained when the scan range increases,the point density decreases. The second test is concerned with the change of accuracy by different scan angle.Seven targets are set and scanned in different angle from 0 to 60 degree but in the same distance to the scanner.In each target's point cloud,the number of points is calculated.The chart of number of points is fitted to the theoretical number of points.The data can be shown when increasing the scan angle,the point density is decreased. The analysis of the results shows that in the different part of a 3D object the point density obtained is different. Therefore,the homogeneous point density on the model is not existed.But according to the research it is feasible to esti- mate the point density and evaluate the level of details,which can be designed for a scanned point cloud on forehand. Thus,a blind scanning can be avoided. Besides the scan angle and scan distance,some other factors also will impact the point density and the resolution of the point cloud,such as the shape,size,materials,environment or surroundings of the scanned object.We do not work on the factors mentioned above in this study.Also,another key factor,the intensity resolution of the point cloud is not in- volved in this paper.