3D Modeling of Urban Facility Using Point Cloud Data for Realization of Smart City
Joon Kyu Park and Keun Wang Lee
(Received April 26, 2020; Accepted June 19, 2020)
Keywords: 3D laser scanner, accuracy analysis, point cloud, shape matching, registration
Among the structures that make up a smart city, tunnels are recognized as an important facility in the US and Canada as an infrastructure for the construction of an eco-friendly city. Total station surveying takes a long time to acquire data and has the disadvantage that the Global Navigation Satellite System (GNSS) needs to be able to receive satellites in tunnel surveying. Recently, 3D laser scanners have been attracting attention as a new technology to compensate for the disadvantages of a small amount of data and a long measurement time of the total station. Three-dimensional laser scanners can be used in various fields. In civil engineering, research and application methods often use 3D laser scanners. Because tunnel surveying is difficult in a GNSS survey due to site conditions, total station surveying has been mainly used. As total station surveying takes a lot of time to acquire data, the use of 3D laser scanners for tunnel surveying has recently been increasing. However, when using a 3D laser scanner, a large number of reference points must be installed inside the tunnel for registration and geo-referencing of acquired data. As a new technology, a shape matching method has been introduced in the field of scanning sensors, and its use is expected. In this study, a tunnel survey using a 3D laser scanner was performed and its utility was evaluated. Also, the accuracy of the shape registration method for recent registration was analyzed. The scanning data was acquired by configuring a traverse using the total station function and compared with 10 checkpoints already installed for accuracy verification. As a result of the accuracy evaluation by comparison with the checkpoints, the maximum error was within 6 cm in the N, E, and H directions, and these results indicate the plane and elevation allowable accuracy of a 1:1000 digital map, suggesting the applicability of the method using the reference point performance and the laser scanning data. Using scan data, not only analysis of the cross section but also continuous analysis of the shape of the scanning section is possible. An additional evaluation of the accuracy of shape matching of the 3D laser scanner data was performed. Using the scanning data acquired five times, geo-referencing of the data acquired at the first station was performed, and then the data obtained was registered by the shape matching method. A maximum of error of 1 m occurred at the end of the data. This indicates that the accuracy of the shape registration can be degraded if the same shape is repeated, such as in a tunnel. Therefore, in tunnel surveying, it is judged that the construction of the data only by shape matching may lower the accuracy. Further research can improve the accuracy of the shape registration method, which will improve the efficiency of tunnel surveying.
Corresponding author: Keun Wang Lee