The Key Technology Study Of Three Dimension Reconstruction Based On Laser Scanning

There is a very high practical significance and application value about thethree-dimensional surface reconstruction by using laser scanning technology. And it is thekey that is how to process the scanning data. The scanning data always is so many that it ishard to directly use. So, the problem which the paper should solve is that how to reducetime with keeping accuracy. This paper made a main research about how to efficientlyreduce the point-cloud data, meanwhile, had a study about triangulation. And the maincontents can be summarized as follows:Firstly, the data which is acquired by LMS100laser scanner can’t be directly used forthe three-dimensional surface reconstruction. The distance data is converted to coordinatevalues through the establishment of a plurality of coordinate system in the chapter2. Inaddition, due to the influence of various factors, many types of errors are contained in theoriginal data. Filter is used to process the original data and the errors of data aresignificantly reduced.Secondly, a new point cloud simplification method has been proposed in chapter3,based on the research for curvature method and uniform grid method. In first step, thefeature points of the scanning line are marked by calculating curvature. In second step, thefeature points which have been marked will be reserved. In the third step, using uniformgrid to reduce point cloud, the points which have not been marked are processed by usinginverse distance weighted average method.Then, considered the point cloud features after reducing, a new point cloudtriangulation method has been proposed in chapter4by the learning the scanning linetriangulation method and the region growing triangulation method. In first stage, the pointcloud data is triangular by using similar scan line triangulation method. Then, in the gridswhich contain marked points, the points are meshed by using region growing triangulationmethod.Finally, through the use of the improved method on the Matlab platform, the paperbuild the system which can complete a three-dimensional image of the coal pile surface by using LMS100laser scanner data.