This study aims to eliminate the labor-intensive target installation process and its associated costs and other issues commonly caused from the current practices of laser based as-built data collection and modeling process. In this study, a laser scan system was utilized with corresponding texture data simultaneously obtained from a digital camera. Based on identified common features in the texture data, an optimized transformation matrix for the point clouds is generated, then the point clouds are registered without using any physical external target. The proposed method was tested at an on-going building construction project. The interrelationships among registration speed, registered accuracy, and size of overlapping area were examined. The field experimental results demonstrate that the proposed target-free geometric data registration method can significantly reduce the registration time without compromising the registration accuracy; thus simplifying and promoting the current laser scanning and registration processes for progressive as-built modeling of construction projects.

No abstract