Reconstraction Of 3D Surface Data Of Quadric From Its Perspective Drawing

Inferring 3D shape of objects in a scene from image remains an important and difficult problem in computer vision. The significance of 3D reconstruction in many projects is more than in theory, and it is also in practice. A number of approaches for inferring 3D shape have been suggested, such as Shape from X, but computer vision is still in its initial stage. The difficulty arises, of course, from the fact that an image is a 2D projection of the scene affected by noise and the process is not invertible without some assumption. As the development of design automatization, manufacturing automatization, terrain survey, vehicle guide, intelligent robots, it is imperative to propose an efficient, exact, robust, general 3D recover system.Based on the perspective drawing, an inverse perspective transform algorithm has been presented and implemented. It can recover 3D surface data of quadric from industry production ID styling drawing or picture.The key idea of our method consists of three modules. The first is image digitalizes, the second is edge detection and link, and the third is inverse perspective transform to recover 3D information of the object. We believe that shape of the 2D contour itself is the most important ones for the shape of the 3D surfaces, so our system is based on image and geometry. We discuss the importance of quadrics in computer aided geometric design, geometric modeling. A new idea for production design has been proposed. It can recover the 3D shape data of quadric automatically from its perspective drawing. A new edge detection and link method has been presented and implemented after we have investigated the key technique and many optimal filters. Extracted geometrical edges from real images are shown. We discuss the method of resolution nonlinear multivariable equations. The relation ship between the root of systems of non linear equations and the solution of the initial valueproblems is discussed. Based on this, a non-linear iterative method that is the same with this paper for solving systems of non-linear multivariable equations is given.Uniform precision analysis model is hard to achieve for large numbers of methods and tools used. Since the explicit uncertainty equation of 3D reconstruction achieved by error propagation theory is not intuitionistic, this paper presents a perturbation analysis model of 3D reconstruction and studies reconstruction precision affected by image digitalization error, matching error and calibration error using multidimensional analysis of statistic method.We have analyzed and implemented the inverse perspective transform theory for the recovery quadrics. Reconstruction of 3D surface of this paper is possible and can be work over wide range of industry production.Experimental results are presented and analyzed for both synthetic and real images. It is shown that such a reconstruction of 3D surface of quadrics is possible and can be work over a wide range of industry production with its high precision and robustness.This thesis is the project supported by Shaanxi National Science Foundation (2002E224) and Aeronautical Science Foundation (03153062).