The operating principles of the photogrammetry algorithms allow you to range from the relief of broad scenarios until the microscopic. Basically, apart from the shooting technique which changes depending on the object and scale you intend to work on, to succeed in the undertaking you need to take perfectly focused photos.
Why perform macroscopic and microscopic scans in the works? For at least a couple of good reasons.
Why perform macroscopic and microscopic scans on the construction site?
Remaining within the scope of building pathologies, for example, it was possible to extend the field of investigation of deep crack patterns. Thanks to the use of recovery strategies supported by artificial intelligence algorithms, in the case of the study of a bridge, it was possible to investigate the morphology of a lesion in the deep layers. The perception of the phenomenon changes at a macroscopic scale.
Furthermore, having the macroscopic and microscopic survey available, it is possible measure with extreme accuracy in space and time the progression of the degenerative phenomenon.
Why use artificial intelligence?
Basically because some of its applications they simplify a lot both the skills required on site by the operator and the post production work.
It is worth remembering that at present, to the best of the authors’ knowledge, there are no lenses capable of capturing frames at high magnification (from 6x to 500x) obtaining a perfect focus. It is a concept well known to photographers specialized in this sector. The best equipment allows good results on partial areas of the frame.
In the research carried out in the shipyard in recent years it has been possible to use the technique of focus stacking. However, those who work on a construction site, between dust, pollution of all kinds, vibrations (which become crucial for those who have to deal with these surveys), are practically impossible to carry out this type of activity.
Let’s also add the lighting conditions and the fact that often, to work without particular interferences it is necessary to perform night scansand the matter becomes really complicated to the point of considering this type of work practically impossible.
Once again, the constant experimentation of new technologies has allowed us to break down many barriersmaking possible what for many colleagues is impracticable and for others, even useless.
Is it possible to perform a 3D scan using photogrammetric techniques at night?
The answer is yes. Lighting technology has made great strides thanks in particular to two specific developments in recent times: that of high performance LED and battery lifeboth governed by an intelligent remote approach.
So to sum up, thanks to artificial intelligence that automates, supervises and governs the acquisition phases on the construction site, thanks to smart lighting devices equipped with AI modules connected to the platform, extremely high-performance and capable of illuminating large portions of the survey scenario, thanks to the evolution of structure from motion algorithms, it is possible to build a new workflow suitable for carrying out this type of work.
Equipment is not enough
The whole relief scenario requires a control room: therefore it’s not enough to have the equipmentbut it is necessary to govern each aspect in a coordinated manner or, as the more or less conscious supporters of linguistic contaminations like to say, “smart”. In essence, leave Deep Learning the control of the acquisition process on the construction site such as the dimming of light sources, that of the real-time spatial position of each element of the scene, the shooting and real-time processing of the frames, the sending of the results to the processing station (on site or thousands of kilometers from the construction site).
Without artificial intelligence it would be impossible to do this workboth due to the prohibitive costs resulting from the presence of specialized personnel, and due to the logistics of the construction sites which do not allow, for safety reasons, this type of activity.
In the authors’ opinion, to achieve the intent of obtaining a digital twin with this approach, it is necessary to simplify the data acquisition phase as much as possible.
