A system scheme of 3D object reconstruction from single 2D graphics based on neural networks

3D reconstruction has turned out to be a significant need in computer aided design (CAD) technology. With the rapid development of virtual reality and industry design, more and more 3D models are required in practical needs. So an artificial neural network based system scheme is presented in this paper, which can be used in 3D object reconstruction from single 2D image. BP neural network (BP NN) is applied in our system due to its non-linear mapping and self-adaptive ability. The 3D objects are represented by Open Inventor, which is a 3D software development kit based on OpenGL. Experimental results show that neural network is a promising approach for reconstruction and representation of 3D objects.