Research On 3D Reconstruction Technology And Algorithm Based On Kinect

Three-dimensional reconstruction is a kind of digitization of computer technology that by collecting the three-dimensional information of entity in the reality world to reproduce the entity with the help of computer technology.As an important branch of multidisciplinary fields such as computer graphics,intelligent information processing,computer vision and artificial intelligence,three-dimensional reconstruction technology has been widely applied in the fields of apparel industry,mass entertainment and reverse engineering and has attracted scholars and researchers' concern.The original scanning equipment was mainly used in the field of project research and was difficult to promote in the market due to its size,high price and complicated operation.The purpose of this paper is to study the 3D reconstruction technology and algorithm based on kinect.The main contents of this paper are as follows:(1)Point cloud segmentation plays an important role in the process of 3D reconstruction.Effective point cloud segmentation can provide more accurate point cloud data for surface reconstruction and feature extraction.However,lower rate of key point matching decreases the accuracy.In order to solve this problem,an improved RANSAC algorithm is proposed.Firstly,a threshold is set to divide the point cloud data into inliers and outliers.Secondly,the Hamming distance is used to perform the initial matching.The angle and the normal vector are set to optimize the segmentation result.In order to ensure that the surface feature of the point cloud to be split is complete and there is no excessive segmentation,then the distance threshold between the two key points is calculated.Finally,an optimal segmentation method is used to get the final segmentation.The experimental results show that the improved RANSAC algorithm is effective and reduces the error rate of segmentation.(2)Point cloud splicing has a wide range of applications in the three-dimensional object reconstruction,the scanning equipment may be limited by the field of view,the object to be covered by itself or the larger size of the object to be measured,so that the scanning equipment cannot obtain all point cloud information of the object under the same angle.The accuracy of traditional ICP is influenced by the initial pose of the cloud,and the robustness is poor.Aiming at this problem,this paper proposes a point cloud stitching algorithm based on the ICP with multi-view cloud data.The axis is combinedwith a threshold.And searching range of the candidate points for constrained threshold is set to reduce the unnecessary calculations.Under the constraint of this range,the nearest point set of Euclidean distance is obtained and the point cloud stitching is carried out by using ICP.The experimental results show that the improved algorithm can reduce iterative time-consuming and improve the splicing accuracy compared with traditional ICP algorithms.(3)The Moving Least Squares(MLS)method is used to observe and process the data and is the final step to complete the 3D reconstruction.First of all,each data in a point cloud must be estimated using a normal estimation algorithm.Secondly,a greedy triangulation algorithm,a mobile cube algorithm,and a Poisson surface reconstruction algorithm are separately introduced.Thirdly,given that the surface reconstruction algorithm is for 3D to reconstruct high-precision requirements,this paper selects the Poisson surface reconstruction algorithm.Finally,through defining the vector field,the Poisson equation is constructed to solve the indicator function,transform the surface reconstruction problem into a solution to the Poisson equation,and the Poisson equation iso-surfaces were extracted to complete the surface reconstruction.The experimental results show that the Poisson surface reconstruction algorithm can complete the task of3 D reconstruction and achieve the desired effect well.