| ICT, non-destructive testing technology, can investigate the internal microstructure,composition, materials and defect status of forgings clearly, accurately and intuitively, bymeans of serial sectioning technique. While, we can't obtain the3D internal defect sizeand location in forgings from serial sections, and don't describe the whole defect. In thispaper, a direct three-dimensional visualization of the internal defect of forgings has beenconducted by means of finite element three-dimensional reconstruction based on serialsections from ICT.Firstly, our method completes the operation of sequence of slice images in thepre-processing stage, including registration, denoise and defect detection. We usetwo-point registration method, median filtering to remove noise. According to defectdetection, we combine edge detection and region growing for recognition of cracks. Basedon the result of edge detection, we select the image binarization threshold, below which allpixels act as seed points for a region growing algorithm.Secondly, Grid-base hexahedral algorithm (GBHA) can reconstruct the regions ofcrack. Our algorithm uses eight-node hexahedral meshes, positions the nodes of the FEmesh at the centers of the existing isotropic voxels, thus, each FE overlaps8voxels in3Dcase. The initial grid structure overlaps the whole solid model, and we use the advancedeven and odd parity rules to remove the exterior elements and the straddling elements. Asurface node projection process is used to match the mesh to the boundary of the solidmodel.Thirdly, this work integrates Laplace, mean curvature and adaptive improved Laplacemethod based on mean curvature, and proposes a new iterative accuracy control method,applying distance field. According to the constraints, we choose one of the three methodsto move a node to the position, which optimizes the quality of the local mesh.Finally, programming at the platform in the VC++6.0development tools andOpenGL to reconstruct the3D internal defect of forgings. |
PDF Full Text Request