Defect Detection Technology Of Solid Rocket Motor Cladding Based On Laser Point Cloud

Laser scanning technology,as a rapidly developing mapping technology,has the advantages of high efficiency,dense point points,high precision,non-contact,especially suitable for defect detection,mapping,geometric measurement,3D visualization and other fields.The specific properties of the point cloud obtained by this technology include important information such as color,normal vector,curvature and coordinate.Therefore,compared with the traditional image-based motor cladding defect detection technology,the defect detection method based on laser point cloud has more important research value.This paper studies the processing,reconstruction and defect detection of coated laser point cloud.The main work is as follows:（1）The coated point cloud data acquisition method is introduced.According to the causes of point cloud noise,it is divided into discrete point noise and hybrid point noise,and the point cloud denoising method based on Kd-tree and bilateral filtering method are used to remove them respectively to solve the problem of point cloud data surface breakpoint because the sensor can’t get effective data at the surface defect of scanning coating layer,this paper adopts cubic spline interpolation method to repair point cloud holes（2）The solid rocket motor is large in size,it needs to carry out multiple rotation scanning of the motor to obtain the surface data of the whole motor cladding layer,so it needs to register and realize the integration of point cloud data.In this paper,the point cloud registration is realized by the ICP（iterative closest point）algorithm.（3）The point cloud data of the cladding layer collected by the laser sensor is ordered point cloud,and the curvature of the defects of the cladding layer varies greatly,so this paper studies the uniform mesh method reduction algorithm based on curvature characteristics,which can achieve the streamlining effect in non-characteristic areas,and can retain complete defect information to improve the defect detection efficiency.（4）After comparing and summarizing the advantages and disadvantages of three different surface reconstruction methods,poisson surface reconstruction method is selected to complete the visualization of three-dimensional surface topography of the cladding layer.The method of point cloud defect segmentation based on curvature feature and Euclidean distance is studied,and the defect segmentation of cladding layer is realized.Finally,the defect quantization is realized.Experiments prove that the defect detection method proposed in this paper can quickly and lossfully detect the inner surface of the coated layer,with high automation and high precision,and meet the requirements of the project.