Automatic Registration of Laser Scanned Color Point Clouds Based on Common Feature Extraction

Point cloud data acquisition with laser scanners provides an effective way for 3D as-built modeling of a construction site. Due to the limited view of a scan, multiple scans are required to cover the whole scene, and a registration process is needed to merge them together. The aim of this paper is to introduce a novel method that automatically registers colored 3D point cloud sets without using targets or any other manual alignment processes. For fully automated point cloud registration without artificial targets or landmarks, this study uses 1) the SpeededUp Robust Features (SURF) algorithm to identify geometric features among the series of scans and 2) plane-toplane matching algorithm to achieve precise registration. For an initial alignment during the registration process, common feature extraction is utilized to perform a 3D rigid-body transformation followed by aligning the view into the reference system. Further alignment is obtained using plane segmentation and matching from the 3D point clouds. The test outcomes show that the method is able to achieve registration accuracy of less than 1o in deviation angle.