Study On The 3D Reconstruction For Point Cloud

Reconstruction the geometry model from the 3D sampling point set is called surface reconstruction. Surface reconstruction is the key point in many fields such as Reverse Engineering, Visualization of medical image, and it is also the hotspot.This paper focuses on surface reconstruction for unorganized points. On the basis of studying the theory of the reconstruction of point cloud, this paper has carried out thorough discussion and analysis on the following parts: Firstly, for the effect of reconstruction, noise points in point cloud has been studied. we analyses the current algorithms of removing noise points ,then adopts an algorithm based on k-d tree to removing noise points based on discussing how to establish the topology of unorganized point cloud . Secondly, we introduce the principle of Power Crust algorithm. The algorithm carries out good results for either uniformly or non-uniformly sample, but it runs for a long time when a magnitude point cloud is reconstructed. In order to improve its speed, a fast surface reconstruction algorithm based on Power Crust is proposed, which has improved the reconstructing speed greatly. Before reconstruction ,non-uniformly samples the point cloud according to the local feature size .With the guarantee that the remaining point cloud are sufficient to reconstruction ,the amount of points used in reconstruction is decreased, and the speed of reconstruction is improved. At the same time, Laplacian smoothing is also introduced. After smoothing process, the feature of meshes can be preserved perfectly. Finally, a surface reconstruction system is developed by VC6.0 and VTK. It has adopted the fast Power Crust algorithm and zero distance algorithm .Surface smoothing is applied. Different reconstruction algorithm can be used according to the characteristic of point cloud.