Evaluating dense 3D surface reconstruction techniques using a metrological approach

This paper discusses an approach for evaluating the accuracy of dense 3D surface reconstruction techniques based on images. So far, the emergence of these novel techniques has not been supported by the definition of an internationally recognized standard which is fundamental for user confidence and market growth. In order to provide an element of reflection and solution to the different communities involved in image-based 3D modelling, we present an approach for the assessment of the metric performance of an open-source set of routines for bundle block adjustment and dense image matching (Apero/MicMac). The evaluation is performed using a metrological approach, through comparisons between image-based 3D generated data and 'reference' data acquired with two hemispherical laser scanners, one total station, and one laser tracker. Aspects of traceability and measurement uncertainty of all these reference 3D data are discussed. The methodology is applied to two case studies, tailored to analyze the software capabilities in dealing with both outdoor and environmentally controlled conditions. Comparative data and accuracy evidence provided by both tests allow the study of some key factors affecting 3D model accuracy.Peer reviewed: YesNRC publication: Ye