Using dense stereo matching techniques in survey

Abstract

In recent years there has been an enormous increase in the dissemination of modelling techniques known as dense stereo matching (DSM) or image-based modelling (IBM) using free, low cost, open source software, especially applications involving the survey of cultural heritage (architecture, archaeology, and town planning). The algorithms used by these software programmes process photographic datasets and provide a 3D model of the scene in question. During complex survey and 3D restitution procedures, these techniques save time (processing) and money: these advantages Can he useful to any professional. This is why it is important to identify the real potential and limits of these software programmes and how they should be used to produce the best possible results. We carried out parallel tests to verify how to use the two software packages: 123D Catch by Autodesk and VisualSfM+CMVS/PMVS2 by Changchang Wu. The photographic dataset was the most critical part of the procedure involving DSM or IBM techniques. The article illustrates three case studies in which only the third study, provides an optimal dataset capable of generating a reliable model. The tests were carried out on three different architectural objects, ranging from a small to large scale; terrestrial laser scansions were used for the metric comparisons of the ensuing models.