Reconstruction Of 3D Objects Utilizing The Frame Features

This thesis studies the reconstruction of 3D objects utilizing the frame features in the scope of Binocular Vision. A reasonable system scheme is proposed, and the kernel functions of the system is implemented and tested. Our experiment results show the feasibility of the proposed system.Our work chooses the Traffic Video Detection System as the research background. Some new algorithms for 3D object reconstruction techniques are presented. The main contributes we provide are:1. On contour extraction algorithm: based on the common used image segmentation algorithms, a new contour extraction algorithm is proposed. Through experiments under the indoor and traffic environment, we show that the algorithm can get a good result.2. On object division algorithm based on 3D information: the algorithms for object surface division are studied and a new algorithm is presented. In our experiments, the expected results have been achieved and the algorithm is proven to be credible. 3D objects reconstruction algorithms based on binocular vision are also studied, and the methods of image surface rendering using OpenGL are analyzed and developed.3. On the system design and implementation: the system of 3D objects reconstruction based on binocular vision is designed and implemented. Using the existing platform of binocular vision in our lab, the integration experiments about our system are also conducted. The experiment results show the feasibility of our 3D reconstruction system.