Research On Workpiece Surface Feature Modeling And Measurement Method Based On Point Cloud

Surface feature modeling and measurement of workpiece are the basic means of product quality control in industrial production.Efficient quality inspection method can shorten the production cycle of products,improve the qualified rate of products,and improve social economic benefits.Traditional contact measurement and detection is inefficient and equipment cost is high,which can’t meet the large-scale industrial product quality detection.With the development of machine vision technology,vision measurement has been more and more widely used.Two-dimensional vision can replace the traditional contact measurement to complete some basic two-dimensional measurement work.However,two-dimensional vision can’t realize the measurement of three-dimensional dimensions such as geometrical tolerance,so a non-contact three-dimensional vision method is developed on this basis.The disordered three-dimensional point cloud is obtained by three-dimensional vision method.How to apply the point cloud technology to industrial production is the main topic of this paper.At present,the basic point cloud processing technology has been developed rapidly,but there are still some problems to be solved in how to apply it to industry successfully,which is mainly reflected in the following aspects:(1)The efficiency of nearest neighbor search of point cloud is low.There is a lot of noise and a large amount of data in the original point cloud.(2)Point cloud normal vector estimation and adjustment is the basis of many advanced algorithms,but for the workpiece point cloud,its accuracy and efficiency need to be further optimized.(3)Most of the classical surface reconstruction methods do not consider the features of the workpiece,which leads to the low precision of the sharp feature reconstruction.(4)The size information of workpiece has been included in the point cloud,but there is no method to extract the size information from it.In view of the above-mentioned problems in the industrial application of point cloud technology,the main research contents of this paper are as follows:(1)For the first two problems.First of all,this paper investigates the different principles of point cloud acquisition methods,through comparative analysis of the advantages and disadvantages of each principle,selects the point cloud acquisition method suitable for the industrial field,and purchases the corresponding three-dimensional scanning equipment.Then,Secondly,the data structure and search algorithm of the point cloud are studied,and the method suitable for the workpiece point cloud is summarized,which provides the efficiency guarantee for the subsequent point cloud processing.Then,the point cloud denoising and downsampling methods are studied,and an adaptive parameter setting method is proposed.Finally,the normal vector estimation method suitable for workpiece point cloud is studied,and an efficient normal vector adjustment algorithm based on breadth first search is proposed.(2)For the problem of surface reconstruction.Firstly,this paper studies the basic theory of surface reconstruction,as well as the classical Greedy projection triangle algorithm and Poisson algorithm,and summarizes their respective advantages and disadvantages and applicable scenarios.Then,according to the surface variation,the workpiece point cloud is divided into smooth features and sharp features.Finally,based on the workpiece features,this paper studies the accuracy and efficiency of surface reconstruction.(3)Aiming at the problem of workpiece measurement.In this paper,the measurement process is divided into two steps:feature extraction and feature fitting,so the measurement of dimension and geometric tolerance has a certain universality.Then,RANSAC algorithm for feature extraction and least square method for feature fitting are studied.Finally,according to the two methods,the measurement method of outer circle diameter and parallelism is proposed.At the last,the hardware and software experimental platform is built,and the proposed method is verified and analyzed.The research work of this paper further promotes the application of point cloud technology in the industrial field,which has higher economic benefits and social value.