Research Of3D Reconstruction And Texture Mapping Based On The Large-scale Point Cloud Data

LiDAR Mapping Technology is a fast, accurate technology to get the spatial dataof real object. The real object of data extraction which based on3D laser scanningequipment is not only simple but high accuracy and the3D point cloud data isobtained by3D laser scanning equipment. It can reflect the real shape of the objectand is widely applied to reverse engineering and virtual environment simulation field.The computer virtual presentation of point cloud data includes the reconstructionof point cloud data and the texture mapping of reconstruction’s model. Reconstructionof point cloud data is mainly to solve the problem of empty point cloud data to getsmooth surface to form a virtual entity of computer vision. Texture mapping of pointcloud model is to make the model vividly and is an important technology in the fieldof virtual reality. Point cloud data which obtained from LiDAR Mapping Technologyis huge and scattered and it makes reconstruction model of point cloud data and thetexture mapping of reconstruction’s model to became a hot topic in computer graphics,spatial information processing, computer vision and computer simulation.This paper studies the Qsplat fast point cloud reconstruction algorithm and usesQsplat algorithm to make the huge number of scattered and unorganized point clouddata together to form a QS file (hierarchy of bounding sphere tree datastructure).Reading QS file directly shorten point cloud reconstruction time duringrendering point cloud data. In general, data texture mapping methods based on theDelaunay triangular mesh are usually computationally slow and the quality ofmapping is low, making them unsuitable for large-scale point cloud data. In this paper,an improved spherical texture mapping method is developed for the reconstruction ofpoint cloud data and it can be implemented based on the Qsplat algorithm. First, theQsplat Algorithm is used to re-establish the model of large-scale point cloud data.Second, the spherical equal-ratio constraint texture mapping is used to obtain themathematical relationship among texture coordinates, sphere and the reconstructedmodel, realizing the spherical texture mapping of large scale point data. Experimentalresults show that the speed and the quality of mapping of the proposed method are much improved compared with those traditional triangular texture mapping methods.