A New Approach Based on Neural Network for a 3D Reconstruction of the Dome of a Bulge Tested Specimen

A three-dimensional reconstruction of the dome formed by a thin hyperelastic specimen during a creep bulge test was carriedout through two different techniques: stereoscopic reconstruction based on epipolar geometry and Digital Image Correla-tion. A suitable experimental device, provided with a sliding crossbar for acquiring dome images so to detect its strain state,was used for the epipolar geometric reconstruction. In 3D reconstruction based on the Digital Image Correlation, the cam-eras/sliding crossbar system was replaced by a different optical system. A new approach to exploit the greater accuracyobtained with Digital Image Correlation, using cheaper techniques, was based on the training of a Convolution Neural Net-work. This training consisted in using a set of points (x, y) of the specimen at different pressure values in order to obtaina heights (z) map of the dome. This approach is aimed to reconstruct the dome providing to the Network thus trained theimages from a single camera placed on the vertical axis of the dome apex.