Research On Simplification And Optimization Of Triangle Meshes Based On Industrial CT Slicing Data

Industrial CT (Industrial Computerized Tomography, ICT), owing to the feature ofbeing able to display the inner structure, composition, material and defect status of theinspected complex objects with2D slicing image or3D mesh model clearly, accurately,intuitively and without any damage to the inspected complex objects, is awarded as thebest non-destructive testing technology nowadays. But due to the limits ofmeasurements and technology, defect status of the inspected objects, and so on, themesh model reconstructed from industrial CT scanning data is often not closed, besides,the reconstructed mesh model with huge data and narrow triangle have the problem ofstorage, transmitting and subsequent processing.To counter the problems above, this paper takes industrial CT scanning data as theresearch object, and focuses on mesh reconstruction, mesh simplification and meshoptimization. The main contents and achievements are listed as follows:①To counter the existing problems of industrial CT measurement, that is, onlyvalid area of the inspected objects is scanned, the contour information about thebeginning and the end of the inspected objects can not be obtained. We present amethod of modifying voxels at end face, this method is benefit for the extraction ofcomplete iso-surface when reconstructing mesh model and the closure of the end face.Besides, an improved MC algorithm is used to reconstruct all voxels and avoid theambiguity of original MC algorithm.②To solve the problem of storage, transmitting3D reconstruction models withhuge data, two triangular mesh simplification algorithms based on edge collapse andtriangle collapse are proposed in this paper. Compared to the majority simplificationalgorithms at present, the presented simplification algorithms work well for generatinghigh quality simplified model, reducing the error of the simplified model whilstpreserving the sharp geometric features of the original model, and making up defect oflosing features of the original model.③In order to repair the hole derived from defect status of the inspected objects,the overlapping of the part of the inspected objects, and so on, a novel hole-fillingmethod is proposed. Firstly, identify all holes existed in the mesh model and performpreprocessing to boundary edges. Then the identified holes are triangulated by theimproved advancing front mesh method. Finally, adjust the position of triangle vertices existed in the patch meshes by the least square fit. The proposed hole-filling approachrestores the original shape and has smooth transition with the surrounding meshes, themesh model is optimized primarily.④After hole filling, triangles in the area of the hole may be spiky, and can notadapt to the surrounding shape of the hole well. Besides, there are some irregularsituations in mesh model, such as narrow triangle, redundant triangle, and so on. Inview of these problems, a normalization method for mesh model is proposed. Firstly, inorder to improve the quality of mesh model, patch mesh and the area of the narrowtriangle are optimized by applying the weighted bi-umbrella operator, and then repairother defects to obtain normalization mesh model.Above-mentioned research results have been integrated into a prototype system ofreverse engineering, this system has been applied successfully in reverse design. Theapplied examples prove that the presented algorithm not only help obtain high qualitysimplified model, but also benefit subsequent processing.