Consistent Point Clouds of Narrow Spaces Using Multiscan Domain Mapping

Abstract

Abstract: Three‐dimensional (3D) range scanning of large spaces, such as civil structures, generates an immense cloud of 3D points with inconsistent data densities due to the limited positions of the stationary scanner, inaccessible surfaces, and narrow pathways. This density variation is the dominant detrimental factor in extracting accurate scanned shapes. This article introduces an effective scan planning methodology for capturing accurate geometry from long and narrow spaces, which minimizes the need for subsequent data approximations. The technique computes an optimum scanning range for each stationary position of the scanner that limits the density variation to a user‐defined value. Three cases are proposed to define the “limited data density” and a FARO®‐LS880 laser scanner is used to illustrate the proposed approach that achieves acceptable scanning results in terms of its critical shape capturing capability, overall point cloud density, and accurate point‐based visualization. The experimental observations confirm that the accuracy of the scanned data can be improved by registering multiple partial scans with restricted density and positioning the data acquisition device close to the critical features. The latter recommended step decreases the incident angle to the world domain, which, in turn, reduces the surface occlusions and data density variations.