Research on 3D objects reconstruction system from 2D input based on GA-BP neural network

With the development of computer aided design technique,3D reconstruction from 2D input has been widely used in the research of computer vision. Plentiful works have been done in three orthographic views, in this paper a new method is introduced under single 2D input. The system in this paper aims at the reconstruction of useful objects in computer design. Angle histogram and corner detector are used in this system to extract feature and size the object respectively, then GA-BP neural network is used to deal with the relationship between the features and 3D objects. This system is developed on the basis of Microsoft Foundation Classes and Open Inventor 5.0 by Mercury. Examples and experimental results are shown in this paper.