| With the continuous and rapid development of the manufacturing industry,the traditional forward direction engineering has been difficult to meet the needs of people,reverse engineering is more and more widely used in new product development and design.Reverse engineering in the new product development and design,not only can greatly shorten the product development cycle,but also can greatly reduce R & D costs.In reverse engineering,the geometric data of object obtained quickly and accurately is one of the most important steps,but with complex inner contour parts,the method of measurement data is limited,from two aspects of measuring precision and cost,this paper presents a method of 3D model reconstruction based on contour imaging,data acquisition is the CCD camera through a connecting piece to installed in the CNC machining center,the lens and LED ring light installation with high resolution,and is equipped with a detachable shading barrel.The work piece is fixed on the worktable of the machine tool after the embedding body is wrapped up,and the high precision cutting is performed by using the numerical control machining center to get the contour image.After the contour image preprocessing,edge detection,sub-pixel subdivision technology and processing of 3D point cloud data measured,and then the point cloud simplification,segmentation and triangulation of the 3D point cloud data,finally 3D reconstruction of the measured object is obtained by curve fitting.The research contents of this paper are as follows:(1)The experimental study of CCD to obtain the fault contour imageThe test platform is built to preprocess the tested parts,selected according to the working parameters of experimental equipment to appropriate shooting conditions and experimental data were collected,the comparison and analysis of experimental results,determine the best conditions for shooting images of high quality.On the basis of determining the shooting conditions,a complete 3D model reconstruction experiment of image acquisition.In this paper,a total of 249 layers of fault contour images are acquired,including a total of 2490 images.(2)Image preprocessing and edge detectionAccording to the collected raw data,study the image preprocessing technology,namely smoothing,gray average method and median filter;then,through the research of the existing edge detection operator,Canny algorithm and Log algorithm,an improved Canny algorithm is proposed in this paper,to a certain extent improve the edge detection accuracy;finally the sub-pixel subdivision algorithm to further improve the accuracy of edge detection,edge detection accuracy compared to the pixel level,the accuracy of sub-pixel subdivision algorithm is improved by 50%.(3)Processing of point cloud dataThe processing of point cloud data is studied,including the simplification of point cloud data,the segmentation of point cloud data,and the triangulation of point cloud data.The first method of streamlining point cloud data,such as the uniform sampling method,bounding box method and grid based method,this paper proposes a point cloud simplification algorithm based on curvature,and the feasibility of this algorithm is verified by an example;finally the triangulation algorithm,and proposes a new point cloud data triangulation algorithm.(4)Surface fitting and 3D reconstructionThis paper mainly studies the surface fitting,including the theoretical basis of surface fitting,NURBS surface fitting and surface fitting.Using IMAGEWARE and UG software based on characteristic curve reconstruction,contour fitting and reconstruction of NURBS surface fitting reconstruction of three different methods on processed 3D point cloud surface fitting,the 3D model of the test piece,through the analysis of the reconstruction error research,NURBS surface fitting reconstruction accuracy is the highest.Finally,the feasibility of improving the reconstruction accuracy is proposed from three aspects of data acquisition,edge detection accuracy and surface reconstruction accuracy. |
PDF Full Text Request