Point cloud quality requirements for Scan-vs-BIM based automated construction progress monitoring

Abstract

Abstract Significant research effort has been focusing on automated construction progress monitoring using the Scan-vs-BIM method. In recent years, various scanning technologies were applied with different success. The general finding is that a higher quality of the point cloud leads to improved monitoring results. Most published works in the relevant area recognise density and accuracy as the main quality parameters of a point cloud. Data quality has been addressed in various ways, by defining arbitrary levels of quality parameters, by evaluating the quality parameters of a point cloud, and by defining parameters of a scanning plan in order to achieve a desired level of quality. However, the relation between the levels of point cloud quality and the success of Scan-vs-BIM element identification is still an open question. This paper presents results of a research in which we defined a more accurate and applicable metric for evaluation of the quality of a point cloud for construction progress monitoring using a Scan-vs-BIM method. The proposed methodology includes the definition of building element classes and the definition of point cloud quality parameters, which have been selected by observing the most significant criteria that influence the success of building element identification. Using a test BIM, around hundred point clouds have been generated with combinations of influencing parameter values. A statistical method was applied to determine the point cloud quality criteria for assuring correct identification of each class of elements. The quality criteria were then validated using three different scanning methods. Results show that the defined quality criteria can be effectively applied in deciding on the appropriate scanning methodology for successful Scan-vs-BIM identification.