Pipe Surface Reconstruction Form Scatter Point Clouds

Recent years, reverse engineering is a research area, which is developing very rapidly. It is one of important technologies in the CAD/CAM field that can shorten the circle of product redesign and manufacturing, and has a wide range of applications. It is the problem of surface reconstruction that is a key element in reverse engineering. With the development of coordinate measuring devices, point cloud data can be easily captured from solid surface. And the sampled point cloud is large, dense, irregular and no topological. The reconstruction of solid models base on the point clouds has become a hot issue, it can be aptly called surface reconstruction from point clouds.In this paper, the research object is the point cloud data captured from pipe surface. The paper proposed two methods to fitting the pipe surface central axis, and the two methods are fitting local cylindrical surface and fitting osculating paraboloid. Then calculate the surface differential geometric properties of points. Thus the scattered point cloud data will be moved to the central axis of the surface. After then the points will be fitted to the B-spline curve using the least square method. Finally making the B-spline curve as the central axis, draw the pipe surface. This way we can reconstruct the pipe surface by modifying the control points.The layout is as follows:The first chapter reviews the study of point cloud significance of surface reconstruction, and the research status of point cloud and surface reconstruction technology. The second chapter introduces some basic concepts and knowledge of differential geometry, including point cloud segmentation, the surface's two basic formula and estimation of its differential properties. In the third chapter, we use the method of fitting local cylindrical surface to generate the points on the central axis. First segment point cloud in accordance with its larger axis span of the data block, and then find its locally best-fitting cylinder for each data block, access one point on cylinder center axis taken as the point on the central axis of the target pipe surface, at last we get sorted points on the central axis of the target pipe surface. The fourth chapter proposes a fitting method for the complex surface, the use of minimum bounding box method which uses a number of small cuboids dividing point cloud data. We set central point as reference point for each small cuboids. Then set up a moving parabolic approximation model for each reference point according to its neighborhood point set. And then project the reference point onto the target surface to get the marked point. Recover the geometric properties of the marked point. Finally move the markers along the normal vector to the center axis. At last we get many unsorted points on central axis of target pipe surface. In the fifth chapter, we introduce a method of fitting the pipe central axis and drawing the pipe surface. First we use the point which got in chapter 4 and chapter 5, to fit a B-spline curve. Then select points on the spline curve and calculate the corresponding tangent, draw space circle using quaternion method. Finally connect all the circles together to form a pipe surface.