The contents of this presentation aim to showcase the choice, use and testing of currently available technologies with the goal of building a smartphone camera-based 3D reconstruction system. The technologies considered for the paper were non-contact technologies of both passive and active variety. The passive methodology was chosen on cost related basis and the availability of libraries integrating the technology.
However, it was proven that passive methodologies were not suitable for use with smartphone camera-based solutions. Thus, Intel RealSense D435i was used as a substitute for the depth map generation. The rest of the device was built with the final cost below the target budget of 100 euro, and with the RealSense camera, a coefficient of variation of 305.478% was achieved with respect to the reference mesh of the scanned Stanford bunny model.
The total reconstruction time achieved was 75.4835 seconds. However, to continue the work, a recommendation to explore active vision reconstruction technologies was made due to the inconsistency of the explored passive method.