T-spline Reconstruction And Pointcloud Registration About Free Form Surface Data

Inspection,reconstruction about the real world object is wildly used in the reverse engineering,manufacturing,geodetic even game industry.In order to interact between the computer terminal and the real measuring data,reconstruct the measured object precisely,the study about the point cloud reconstruction and registration becomes the hotspot in recent year.In this paper,the research on point cloud reconstruction and point cloud matching is studied in response to the actual need for a robust and high-precision matching and reconstruction scheme in the process of multi-view free surface measurement,analysis and evaluation.a study of point cloud reconstruction based on T-spline and point cloud registration with overlapping areas is conducted,main tasks as follows:(1)The tensor product-based skining algorithm is a commonly used configuration algorithm in engineering practice,but the control point topology grid of the tensor product surface is limited to a fixed quadrilateral rectangular grid,which is likely to cause the difficulty of the stitching and cropping in practical applications..T-spline is a flexible configuration algorithm for topological grid of control points,so T-spline has the advantages of easy splicing and cutting.At the same time,T-spline surface has fewer control points than tensor product surface,which is beneficial to improve the efficiency of calculation.Therefore,this paper will study the skining algorithm based on T-spline,it improve the step of fitting the control curve in the current T-spline skinning algorithm,it will not fall into an endless loop like the current algorithm.And also improved stability of the overall algorithm.(2)In order to solve the problem that the source point cloud and the target point cloud partially overlap in the actual matching,and the characteristics of the measured object are not obvious,this paper proposes a registration scheme based on region division that follows the "coarse-fine" process.This scheme makes great use of the prior knowledge of overlapping regions,and improves the stability and robustness of the matching scheme.In rough registration,ISS + High-Dimensional Histogram(FPFH)is used for rough registration,which makes full use of the point cloud information and avoids the problem of unobvious features.Sparse registration is used in fine registration to solve the interference caused by outliers on registration.The SPARSE-MICP algorithm is proposed to make up for the shortcomings of the sparse ICP,such as easy to fall into the local optimal,and the sparse TDM slip.(3)An experimental study based on the T-spline skinning algorithm was carried out.The aero blades were used as experimental objects to reconstruct the point cloud data in different regions.The results show that compared with tensor product,T spline surface can greatly reduce the number of control points of surface configuration under the condition of the same fitting accuracy.A verification experiment was carried out for the point cloud matching scheme proposed in this paper.The experiment showed that the scheme is effective for the problem of overlapping regions and the characteristics of the measured object are not obvious.According to the precise registration algorithm SPARSE-MICP proposed in this paper,compared with a variety of fine registration algorithms,the experiment shows that SPARSE-MICP has higher accuracy,stability and robustness to different levels of noise,outliers and initial positions.