Adaptive And Quality Hexahedral Meshing From Industrial Computed Tomography Image Data

The 3D reconstruction technology based on ICT(Industrial Computed Tomography)image has become a research hotspot with the development of Industrial Computerized Tomography technology in the application of high-precision measurement and non-destructive testing technology.Although there has been tremendous progress in the area of 3D geometric modelling,it remains a challenge to generate 3D meshes directly from imaging data.In this paper,we extract hexahedral meshes directly from industrial computed tomography images data.The meshes here remain adaptive and quality in the extraction process.The main work of this thesis is as follows:1.We proposed an image-denoising method based on both wavelet transform and guided filtering.This method is able to tackle noises generated in the process of image acquisition and therefore avoid noise impact from reconstruction results by keeping edge information of the original image while removing the noise.The simulation results show that this new imagedenoising algorithm is more effective than others.2.In order to rapidly extract uniform hexes meshes from industrial computed tomography images data,we use a top-down octree subdivision coupled with the improved dual contouring method.By this means,uniform hexahedral grids are generated with better face sensitive feature and correct topology.3.However,for a given precision requirement,uniform meshes are always over-sampled with unnecessary small elements;therefore,adaptive meshes are preferable.In this part,we introduce a template-based approach for uniform hexahedral 2-refinement.Meanwhile,we control the level adaptively with an error function and a tolerance.Also the ED-error and a trilinear function is defined to calculate the vertices from every adaptive hexahedral mesh using the corresponding templates.The result shows that adaptive meshes can provide a minimal number of elements while preserving important features on boundary surfaces for efficient and accurate calculations.4.To improve the poor boundary hexahedral mesh,we extend the guided filter to mesh smoothing,which is computationally efficient while keeping the edge features.Therefore,we describe an approach based on the guided filter to achieve the purpose of improving the quality of hexahedral mesh.Finally,the scaled Jacobian is chosen as our corresponding quality metrics to measure the quality of hexahedral meshes,showing that our method can be used to improve hexes elements as well as maintaining the sharp features.