The Research Of Compression And Registration Based On Point Cloud Of Terrestrial 3D Laser Scanning

With the rapid development of science-technology and economy,three-dimensional laser scanning technology has made great progress in various application areas,which makes three-dimensional laser scanning technology has widely applied in reverse engineering,deformation monitoring and medical field.The three-dimensional laser scanning technology which is not in contact with the scanned object under the condition of surface feature and texture feature information can get the object,then the scanning point cloud datas are obtained by a series of data processing,we can get visual objects features complete,therefore,the development of three-dimensional laser scanning technology plays an important role in the three-dimensional modeling of objects.However,the three-dimensional laser scanning technology is not tentative enough,there are still many urgent problems in their own scanning system and data processing,especially for the precision of multi station cloud data registration is an important factor which restricts the accuracy of the data processing and three-dimensional point cloud modeling,As when the three-dimensional laser scanner observes the scanned objects,which is often necessary to assume the complete feature information of the scanned objects from multiple stations,but these stations are in different coordinate systems,so the coordinate system needs to be translated and rotated in order to obtain the complete information of the target object,so that the scanning results in different coordinate systems can be unified into the same coordinate system.Therefore,how to integrate the stations with high quality and high efficiency in different coordinate systems to the same coordinate system,so which need to be fine and efficient research on point cloud registration algorithm,so as to improve the accuracy and efficiency of point cloud registration.Therefore,the paper focuses on the research of three-dimensional laser point cloud data compression and registration algorithm.The following is the main work of this paper.(1)The paper summarizes the research status of the compression algorithm and registration algorithm of three-dimensional laser scanning technology both at home and abroad,and introduces the basic principle,classification,application field and characteristics of three-dimensional laser scanning technology,as well as the process of point cloud data processing(2)The amount of point clouds data used in this paper is relatively large,and the data is relatively scattered,so the regional center of gravity compression algorithm in different proportions for point cloud data compression,then analyzing the characteristics and quality of comparison after the compression of point cloud data in line with the different compression ratio,the paper use VS2013 as a platform to achieve regional center method of compression test.(3)Through the study of point cloud registration method,introduces the commonly used coarse registration algorithm and fine registration algorithm of ICP,then the article introduces the structure of k-d tree algorithm for point cloud data and the nearest point search,meanwhile using BBF(Best Bin First)algorithm improves the closest point search algorithm of k-d tree,and then using the algorithm of k-d tree to accelerate ICP algorithm,and comparing to the classical ICP algorithm and ICP algorithms which is accelerated by k-d tree algorithm through the VS2013 platform,the experimental results show that the accelerated ICP algorithm using k-d tree,not only improves the efficiency of point cloud registration,but also improve the accuracy of registration.