Skeleton-based 3D reconstruction of as-built pipelines from laser-scan data

• This study proposed an automated approach to generate the as-built 3D pipeline model.
• The basic principle is that a pipeline is defined by its skeleton.
• Two-step process is used to extract skeleton from the noisy and incomplete data.
• Performance of the approach was evaluated by field experiments at an oil refinery.
• The approach unlocks the potential to reconstruct the entire as-built 3D facility.

There has been a growing demand for an as-built 3D pipeline model. Although several studies on automation of the process of reconstruction of as-built 3D pipelines have been carried out, previous approaches have been limited to the generation of only a portion of an entire 3D pipeline. The aim of this study was to propose an automated approach to the generation of as-built 3D pipeline models of entire pipelines composed of straight pipes, elbows, and tee pipes from laser-scan data. First, the skeletons of individual pipelines are extracted. Then the extracted skeletons are segmented into their individual components, and a set of parameters for them are calculated. The experimental results show that the proposed approach is robust to incompleteness of the laser-scan data, as well as to noise and to density variations in the data. As a result, the proposed method enables the generation of reliable as-built 3D pipeline models.