A Study Of The Simplification Algorithm Of Point Cloud Data Based On The Image Matching Technique

With the "smart city" proposing,the higher requirement of 3D modeling technology is the necessary prerequisite to promote the development of "smart city".Although the laser three-dimensional scanning technology already have existed and been used widely,there are also is much inconvenience in getting data of urban areas and the establishment of real three-dimensional scenes.With the UAV platform updating fastly,aerial photogrammetry technology has developed rapidly.Oblique photogrammetry technology has features such as getting the data efficiently,being effected small by the environment 5 costting low,getting the three-dimensional model veritably and information-rich,etc.It has been widely used in the fields of land management and planning,agriculture,forestry,archeology and so on,playing a huge role in promoting the development of "smart city".The key technology of obtaining 3D point cloud data in the oblique photogrammetric system is multi-view image matching technology,with wealth of surface texture information and strong visual visibility,it give a great help to improve the efficiency and reduce cost of the three-dimensional model surface's reconstruction.However,the image matching point cloud data is very huge,this provides a great inconvenience to the late data processing,management and so on.Consequently,how to simplify the image matching point cloud data and with a minimum of points to better present the three-dimensional model is the paper's research content.This paper first introduces the basic principle of oblique photogrammetry technology and the basic flow of obtaining image matching point cloud data.And then through the Photoscan software to get the initial point cloud data,after pre-processing,I get the experimental data which need to simplify.The main research contents:For the experimental data,there are three kinds of classical calculations to simplify point cloud,including random sampling method,curvature sampling method and uniform grid method.Then respectively compare the effect of point cloud with the three methods in different reduce rate,analyzing the shortcomings of the three methods and proposing an improved algorithm that is the combination of two calculations,The basis idea is that firstly construct the triangular grid,and then find the triangular grid normal vector,and according to the triangular grid vector to get the normal vector of each point,acquire the normal vector angle between the adjacent point,than in the light of the angle gets a threshold to simplify the point cloud.For the adiacent points if the angles between them are greater than the threshold,keep retaining,otherwise less than the threshold keep equally sampling until reach the demand reduction rate.Finally,evaluating the improved algorithm on the surface and volume and then analysising the 3D deviation.It is proved that the improved simplify point algorithm is better than the first three traditional methods.