Applications for 3D imaging systems (e.g., laser scanners) have increased significantly in the past decade. Data from such systems often require registration. Spheres are often used for registration because they look the same from each direction and provide well defined reference points: fitted sphere centers. Common practice is to first scan individual targets at high density for high accuracy of fitted centers and then scan the entire scene at lower density. This, however, prevents the use of 3D imaging systems in a fully automated environment. It would thus be desirable to be able to determine the sphere centers from the low density scan. In this paper, we investigate how different scanning densities affect the locations of the fitted sphere centers. Analyses of high and low density scans show that high density scans are not necessary because the sphere centers can be estimated from sparse data without significant loss of accuracy.
Proceedings Title: International Symposium on Automation and Robotics in Construction | 25 | | ISARC'08
Conference Dates: June 27-29, 2008
Conference Location: Vilnius, Lithuania, USA
Conference Title: International Symposium on Automation and Robotics in Construction
Pub Type: Conferences
3D imaging systems, target based registration, sphere fitting, construction automation