Algorithm And Application Of Caves Three-dimensional Laser Point Cloud Automatic Stitching On Unmarked Point

With the rapid development and application of Terrestrial 3D laser scanner in the field of three-dimensional modeling, currently, the processing of the terrestrial 3D laser data point cloud is a research hotspot and difficulty.In the point cloud data post-processing,the crucial step is to stitch multiple local scanning point cloud for a complete target by certain methods, then do different point cloud post-processing.Due to the working principle of laser light travels in straight lines, normally, several sites will be needed to gain the target object complete information,Then eliminate point cloud redundant data, 3-D modeling, post-processing and so on.In view of the KARST tourist cave?s special economic value,the terrestrial laser scanner used in three-dimensional modeling in cave is very favorable to promote the development of cave tourism industry,this paper studied the automatic precise splicing method that based oncaves 3D laser point cloud data without landmarks.First of all,pretreatment the cave point cloud data, including adopt based on filter algorithm statistical analysis to filter out the noiseof the cave point cloud data,And usethree-dimensional Voxel reduction to simplify the huge amounts point cloud data which filtered out noise points, using three-dimensional Voxel reduction could streamlining the point cloud number at the same time furthest keep original geometric feature of the point cloud data.Then splice the streamlined point cloud data,The splice algorithm includ:Based on three-dimensional point cloud multiple cave SHOT local reference descriptors of the same name the initial feature points seamless splicing and based on3D-NDT cave point cloud automatic precise registration algorithm.Multiple cave point cloud seamless initial splice algorithm based on homonymous feature points mainly aimed at multiple scanning point cloud with closing conditions.First step adopt local reference 3D SHOT reference descriptor to get homonymous feature points between two scan station, Second step in the process of sequence matching the errors are accumulated,inorder to eliminate them, this paper distribute the cumulative splicing errorto each two scanning point cloud splicing parameter based on the closed constraints, Experimental analysis shows that this algorithm can achieve seamless initial joining together, thus can more objectively reflect the integrity of the target;Based on unmarked points 3D-NDT cave accurate automatic set-point stitchingalgorithm is that turn point cloud within a three-dimensional Voxel data into a continuously differentiable probability distribution function, Using standard optimization techniques to determine the optimal matching between two point cloud,and the stitching process doesn't need to use character is tics calculation and matching of the corresponding points.The study results show that this algorithm in splicing accuracy and splicing time than other stitching algorithm is predominant,The two-two scanning point cloud best stitching accuracy is 1.2cm, he scanning accuracy of the overall mosaic is 2.09 cm,For the latter part of the cave precision three-dimensional modeling to provide a reliable data.