Simplification And3D Surface Reconstruction For Point Cloud Data

The traditional top-down engineering design needs high demand for the designers’experience. As the increasingly complexity of product models, the shortcoming having a lowefficiency is more and more prominent. People pay more attention to Reverse Engineeringbecause of its short manufacturing cycle, low cost and high efficiency. Due to the improvedaccuracy of three-dimensional scanner, point cloud data models obtained are complex andhave huge amount of information，therefore, as the two key technology areas of ReverseEngineering, simplification and surface reconstruction have always been the importantresearch question for scholars. This text further researches the processing algorithm inReverse engineering, realizes point cloud simplification, respectively realizes triangular meshreconstruction and NURBS surface reconstruction for point cloud simplified. Relying onalgorithms in this text, develop a simple, application of strong and effective point cloud dataprocessing software for small-scale and large-scale data.The distribution of point cloud data obtained by three-dimensional scanner has a certainregularity that points distribute more densely near the scan line. So, in order to prevent wasteof memory resources causing by processing point cloud data directly, in this paper for thecharacteristics of scan line point cloud, first, we conduct a pre-processing for data. First of all,we layer the point data and remove the noise points that might impact post-processing, afterthat, sort the point cloud data on the basis of layer.Based on the local feature of B-spline curve, this paper proposes a simplificationalgorithm for point clouds which takes B-spline curve as standard based on retainingboundary characteristics. In this paper, the algorithm combining with the existing featureextraction method has a good simplification result for the complex models and large-scalepoint cloud data models, which can remain the detail feature of physical models.Experimental results further show that the algorithm is simple and validated.This article introduces some related knowledge of triangle, and Delaunay triangulationmethod is described. Based on Delaunay triangulation, this text proposes a triangulationmethod for three-dimensional point cloud data, which is good for scan line point cloud. Andthen we apply the method to point clouds simplified. The experimental results are right; thisalgorithm has a fast formation for triangular mesh and a clear result.Finally, we realize NURBS surface reconstruction for point cloud simplified. First, somecommon block methods for point cloud data are introduced, and then block approach in thispaper is elicited. After that we introduce some basic surface theory and describe some main surface reconstruction methods, at last elicit the surface reconstruction method in this paper.After blocking the point cloud, the points in a block are used as data points for surfacereconstruction. This text uses NURBS surface reconstruction technique to reconstruct surface.Reconstruction algorithm is based on the definition of NURBS surface, which is, conductingtwice the inverse calculation of NURBS curves. Finally, control points for NURBS surfacesare obtained. We fit the control points with NURBS surface and get NURBS surface patches,finally, stretch the NURBS surface patches smoothly and get final3D model.In this paper, this software is designed in VS2008development platform, using c++language, under the MFC framework and applying OpenGL technology. In support ofOpenGL library we can easily achieve the display of point cloud data, view transformation,size changing, drawing the triangular mesh surface and post surface models. The procedureuses the dynamic link library technology to increase the extensibility and uses themulti-threading technology to make experimental result easier to observe.