Unorganized Discrete Surface Co-Registration

Unorganized discrete surface matching is one of the hot issues of current research, which is also the key technology of the process of 3D point cloud data obtained by laser scanning technology. Based on the rigid body transformation model, DEM matching without control points utilizes whole surface information to solve their spatial differences. Considering the progress of 3D point cloud derived from laser scanning technology, two main problems are studied in depth in this research. One is to convert the coordinate system of the unorganized discrete surface into the real geographic coordinate system. And the other is to merge the point clouds, which are acquired as block or strip for the limitation of the data acquisition technology and/or equipment.Firstly, based on the least square matching technology, a novel Least Point to Grid (LPG) algorithm is given for matching the unorganized discrete surface and the gridded DEM. Associated with the proposed algorithm, the unorganized discrete data set and gridded DEM can be co-registered, therefore, the unorganized discrete data will be changed into real geographic coordinate system. LPG algorithm takes gridded DEM as the reference surface, and then resolves the matching parameters using the criterion of shortest vertical distance. The local plane fitting method with restrictions conditions is used to calculate the gradient of each point on the discrete surface. In addition, a point searching method based on 2D network structure is adopted to accelerate the adjacent point searching, which is frequently processed during the plane fitting. The experiments based on simulated data sets are performed to test the proposed LPG algorithm. The experimental results show that the accuracy of LPG is 10% higher than that of ICP, and the efficiency of LPG is 54% higher than that of ICP.Secondly, this algorithm has been improved based on the LPG algorithm to achieve unorganized discrete surface matching,which modify the reference surface into unorganized discrete surface, called Least Point to Point(LPP)algorithm.The splicing problems between neighboring point cloud data could be solved by this algorithm. Inverse distance weighted (IDW)is used to interpolate the corresponding point coordinate from discrete surface. Simulation results show that unorganized discrete surface matching can be correctly completed by LPP algorithm. The accuracy of LPP is comparable to ICP, however, the iterative convergence and efficiency are that of significantly improved, and the efficiency is improved by about 28%.Correlation algorithm has been achieved through the program, and also developed a corresponding functions software. A technical way has been provided, which solves the unorganized discrete surface problems and to incorporate them into the geographic coordinate system. The process of regularization of data has been eliminated by the algorithm. The point cloud data generally have massive features, therefore higher efficient data processing algorithm is very important, research results have certain reference value for real applications.