| Industrial computed tomography(ICT)technology,owing to the feature of being able to obtain data from inside and outside of workpiece and without any destruction to objects,has become one of the popular methods of 3D data acquisition in reverse engineering design.However,there must be noises and shallow geometric features due to some factors,such as,environment,control system accuracy,measurement method,data missing,and reconstruction algorithms during data measurement and transmission.In this case,the quality of the reconstructed mesh based on volume data of ICT is not high,such as,big iso-surface error,high noise,and poor triangle quality,which directly affects the accuracy,efficiency,computational convergence of finite element analysis,and the efficiency and stability of rapid prototyping.In view of the above problems,this paper focuses on several key technologies that effect the quality of the renconstructed mesh based on volume data of ICT,such as,the preprocessing of ICT images,the high precision extraction of mesh iso-surface,mesh smoothing,mesh simplification and mesh standardization.The main contributions of the paper are as follows:(1)2D Otsu method is improved to enhance the features and boundaries of ICT images,and distributed estimation algorithm is introduced to accelate the 2D Otsu solution.Firstly,the guided filtering with better feature retention ability is employed to denoise ICT images.Secondly,the neighboring image information template in the 2D Otsu algorithm is improved,and the target probability and background probability in the image are accurately calculated.Then,a distributed estimation algorithm with strong global search capability and fast convergence rate is used to find the optimal threshold.Finally,the optimal threshold of 2D Otsu is adopted to intensify the edge details of ICT images.This method effectively enhances the edge details of CT images.(2)Aiming at the deficiency that the accuracy of iso-surface extraction is low on account of noises and local grayscale changes in ICT images,marching cube algorithm is modified to extract the iso-surface accurately.Firstly,methods of gradient amplitude histogram and the minimum interclass variance are employed to construct the self-adaptive iso-value field of ICT volume data.Secondly,the threshold of each voxel vertex is calculated by the trilinear interpolation,and the position and normal of the intersection point between boundary voxel and the iso-surface are determined.Then,according to the longitudinal size of the workpiece,an adaptive method for repairing end-face to improve the accuracy of the iso-surface is proposed.Our method effectively improves the accuracy of the iso-surface via using an adaptive iso-value field instead of a single global threshold.(3)In allusion to the challenge of mesh smoothing while preserving the features of mesh,a bi-normal smoothing method based on the feature of mesh vertex by combining the facet normal reflecting global geometric change of the mesh and the vertex normal reflecting local detail of the mesh is proposed.Firstly,the tensor voting theory is employed to classify the vertices of the mesh.Secondly,in order to obtain an accurate facet normal,we construct an accurate guided normal after the guided filtering is extended to the field of geometric processing.Then,compute the vertex normal accurately according to the fitting of piecewise smoothing patches from feature regions.Finally,the vertices are iteratively updated by using facet normal or vertex normal based on vertex features.Our method can remove the noises while better preserve the details of the mesh,and the surface error of the smoothed mesh is small.(4)In order to improve the quality of the reconstructed triangular mesh,a mesh standardization method based on particle swarm optimization(PSO)is presented.Firstly,each vertex of the mesh and its neighborhood vertices are fitted to a cubic surface by using the least-squares method.Secondly,the modified PSO is employed to find the optimal of vertices.Where,the fitted local surface is considered as the region constraint of the PSO,and the optimal average quality of the local triangles is regarded as the target of the PSO.Finally,we regulate the vertex according to the normal angle between the original vertex and the optimal vertex,which can ensure the details of the mesh.In order to avoid falling into the local optimum and speeding up the convergence speed,the PSO is improved by introducing the central particle,the constraint factor and the adaptive inertia factor.Compared with the existing standardization methods,the proposed method can effectively improve the triangle quality of the mesh while retaining the details of the mesh.(5)Most of the existing mesh simplification algorithms cannot take into account reduction ratio,the details,and the quality of triangle in the meantime.In order to resolve the deficiency,an adaptive simplification and optimization method for mesh based on dynamic error and PSO is put forward.Firstly,the projection estimation method is employed to calculate the coordinate of the collapsing vertex,both dynamic thresholds including distance error and angle error are utilized to compute the collapsing cost,and the mesh is simplified according to the cost.Secondly,PSO is improved to standardize the narrow and long triangles of the mesh.Compared with the existing methods,the proposed method can effectively control reduction ratio and result in high-quality simplified mesh while preserving the details. |
PDF Full Text Request