Construction of three-dimensional models of mechanical products from their orthographic views using computer vision

Abstract

The construction of a three-dimensional (3D) model of objects is a key element of many interesting applications such as computer graphics and computer-aided design (CAD). Although many techniques for the construction of 3D models are available, these techniques are still the focus of attention from researchers in the computer vision community. In the current paper, a low-cost vision system has been employed to construct 3D models of small mechanical products using the image-based modelling technique and the basic principles of the orthographic views. The system has been calibrated to compensate for the variations in size of the captured images resulting from the difference in dimensions of the product to be reconstructed. The 3D models are constructed from two captured images, which represent two different views of the product. The constructed 3D model can be viewed through a developed software or exported to any CAD software, such as AutoCAD, for further editing. The mathematical model of the employed 3D construction technique is presented and the verification of the proposed system is introduced.