Analysis And Research On Accuracy Of 3D Point Cloud Data Processing In Reverse Engineering

With the rapid development of reverse engineering technology,the technology has become more and more influential in the field of machinery.Point cloud data processing is a key part of reverse engineering technology.The accuracy of point cloud data has a direct impact on the accuracy of subsequent three-dimensional reconstruction models.In recent years,studying the factors that affect the accuracy of point cloud data and the solutions are the major research directions of the reverse engineering profession.This article takes the actual project as the background and the research content is mainly divided into the following three parts:The first part studies the factors that affect the accuracy of point cloud data collection.Because the point cloud data is acquired by a three-dimensional laser scanning system,the system error directly affects the accuracy of the three-dimensional point cloud data.First of all,the single-edge method and the cross-bilateral method were used to calibrate the three-dimensional laser scanner,and the system error of the three-dimensional laser scanner was detected.Then the effect of system parameters and the properties of the measured target on the accuracy of point cloud data collection was studied by using the control variable method.Through the combination of theory and practice,the specific influence of the external environment on the accuracy of point cloud data collection is analyzed.Finally,the concrete methods to improve the accuracy of point cloud acquisition are proposed from two aspects of subjective factors and objective factors.The second part mainly analyzes the mosaic accuracy of multi-viewpoint clouds through experiments,and studies several point cloud noise reduction techniques.In practical engineering applications,a three-dimensional laser scanning system generally cannot directly obtain all the point cloud information of a target to be measured from a single station,and requires multiple scans to acquire multi-viewpoint cloud data.Therefore,the multi-viewpoint cloud data needs to be stitched.The most commonly used methods for multi-view cloud mosaicing are based on target mosaic technology and cloud-based mosaic technology.Through the experimental design,the effects of target number,placement method and numberof target ball scanning points on the precision of the target splicing technology are studied;the effect of the average sub-sample point distance and point cloud overlap rate on the accuracy of the inter-cloud splicing technology.By comparing the stitching time and the stitching error,the advantages and disadvantages of the two stitching techniques are analyzed.In the third part,the application of point cloud data in reverse engineering is studied with an example of a water injection system in the project.Firstly,the point cloud data is acquired by a three-dimensional laser scanning system,and then splicing and noise reduction processing is performed.Then,a three-dimensional model is created based on the point cloud data using a three-dimensional reconstruction technique,and the accuracy of the three-dimensional model is analyzed.