Research On Partition Registration Method Of 3D Laser Point Cloud

Three-dimensional laser scanning technology has been developing rapidly in recent years, it has many advantages, such as fast scanning speed, abundant data information, high-density, high-precision and so on, therefore, it has been widely used in surveying and mapping and other industries. The approach to increase the registration accuracy of three-dimensional laser scanning point cloud data has always been the focus and difficulty in the technology. There are many algorithms about point cloud registration, further studies has been made into the research on registration method by scholars from different point of view as well, and it actually improved registration accuracy and efficiency of the method or algorithm, promoted effective applications of three-dimensional laser scanning technology in various fields.This paper introduced the basic ideas, mainly principles of three-dimensional laser scanning point cloud technology registration and commonly used algorithms, summarized and analyzed several common coarse registration method which can provide a good initial value for fine registration and provide conditions for the subsequent two stations point cloud clipping the overlapping area; through a test we determine the three-dimensional terrestrial laser scanning point cloud data relative accuracy, taking remote sensing of the Chinese Academy of Sciences building, the Ancient City site of Gaochang, the Ancient City site of Beiting for example, we made a reasonable scanning scheme including the site layout, parameter setting, working time these three aspects, completed high-precision, high-density 3D point cloud data and high-resolution image collection in experimental areas. To explore a new method of point cloud registration, registration is carried out after partitioning and cutting the overlapping area point cloud of adjacent sites in different proportions used the partition local registration method, then realized point cloud coloration and 3D modeling in Riscan pro(a point cloud processing software) based on registered point cloud data and images of various experimental areas.The main research contents and innovative achievements are as follows:1. Analyzed and obtained relative accuracy of the terrestrial three-dimensional laser scanner we used, i.e. at 72" resolution from the scanning center about 100m, the accuracy error is less than 5mm in point cloud, which provided data support for the subsequent research on registration approach of point cloud.2. Made two different cutting programs based on the coarse registration, and took original point cloud of the overlapping area as a reference, completed fine registration of point cloud of the overlapping area in different proportions and obtained its error results at the same time. Using comparative analysis, we got the best partition and cutting sizes.3. Using the above method to complete the fine registration of the point cloud data for all sites in the experimental area, experiments show that it can improve registration accuracy, and its error is consistent with the random error distribution, the precision of effects and visual effects can meet the requirements. Completed point cloud coloration and 3D modeling of experimental areas, which can prove the method of partition registration proposed reliable and applicable according to practical application.