| 3D shape registration is a crucial problem on research of reverse engineering. It is a significant research direction in the field of computer vision, graphics and image processing recent years as well. Furthermore, it has an extensive and urging application in the fields of CAD, Virtual Reality, Plane Manufacturing industry, Medicine and so on. The section, however, is difficult in reverse engineering because of the complexity and messiness of point-clouds. This paper deeply researches the mark points graphics and image processing, 3D recognition, point-clouds registration, point-clouds optimizing processing. 3D registration system by using mark points and automatically registering multiple views is developed based on the least-square method and 3D recognition in order to output the surface model that can be used by CAD/CAM software directly.In the paper, chapter 1 presents the application background, significance of engineering and registration technique. Chapter 2 analyses the theory of raster-scanning, the hardware and several registration methods, puts forward particular project by using mark points and designs structure. In chapter 3, the methods on how to process mark points image are explored deeply. Firstly, the pretreatment of mark points image is presented, and the algorithm on ellipse estimating by using iterative least-square method is proposed and developed in order to process accurately. Finally, the point-clouds filtering algorithm is discussed. Chapter 4 discusses the registration process, the principle of delaminating data is proposed. A multiple views registration algorithm is proposed based on delaminating data, automatically searching and least-square method following the research of registration algorithms and some theories, samples and data are given. Meanwhile the theory of optimizing point-clouds is discussed. Chapter 5 introduces the basic content of output data format and realization of it. The last chapter draws a conclusion. |
PDF Full Text Request