Research On Spatial Registration For Three-dimensional Laser Scanning Data

Three-dimensional spatial information is one of the hot issues of scientific research in the field of Surveying, how to accurately obtain the three-dimensional spatial information is the basis for the study of spatial information. Appearing of three-dimensional terrestrial laser scanningtechnology for three-dimensional spatial information acquisition provides an important means of measurement.Currently,three dimensional terrestrial laser scanning hardware has become perfect,but a massive point cloud data acquisition process has no perfect solutions,there are still some needs to be further researched urgently in the key technology,this paper is researching point cloud registration,which is a key step in the point cloud processing.In this paper,introduced the basic principle of terrestrial laser scanning system, analyzed the causes of errors in point clouds,and finally carried out three studies for the threedimensional point cloud data registration method. Firstly,in this paper,considering the feature of building surface point cloud,presented a mode of point cloud registration based on point and face geometric features.Confirmed by experiment,point cloud registration algorithm,based on the feature of points and face,can effectively achieve resisting surface. Comparing with point cloud registration algorithm,only used the feature of points,this algorithm can achieve higher accuracy;secondly,on the base of the traditional ICP registration algorithm,improved this registration algorithm from three aspects of the control point selection,the search space and characteristic measure. improved ICP algorithm can achieve point cloud registration,compared to conventional ICP algorithm,this algorithm reduced the computation time,reduced the number of iterations,improved the accuracy of point cloud registration;finally,by analyzing the rule of multi-site cloud registration error propagation,satisfy the closing conditions for multi-site cloud,the paper gave a whole point cloud registration method. the whole point cloud registration algorithm can be reasonably allocated closure difference between point cloud the seamless.Confirmed by experiment,the registration algorithm based on the geometric feature of points,the registration algorithm based on the geometric feature of point and face,the improved ICP registration algorithm and the whole point cloud registration algorithm that can be used point cloud registration under geographic scene,and achieved more site clouds seamless.