China
Africa
Explore chapters and articles related to this topic
Computer-Aided Design of Additive Manufacturing Components
Published in Linkan Bian, Nima Shamsaei, John M. Usher, Laser-Based Additive Manufacturing of Metal Parts, 2017
A vertex-based STL file offsetting technique is described by Koc et al. (Koc and Lee 2002). In the method, the weighted average vertex normal vector is calculated from the normal vector of the vertex sharing triangles from the STL file. Each of the vertices from the STL file are then translated toward the weighted normal direction with a user-defined distance or shell thickness. Self-intersections, loops, and irregularities are determined and eliminated by identifying their loop direction to ensure valid offset contour of the 3D model. The thickness or offset distance of the shell is considered as user-defined parameter although it can be an important parameter for self-supporting hollow object. Alexander and Dutta (2000) considered the shell thickness as the design parameter and proposed adaptive or localized wall-thickening technique. They measure wall constructability property (WCP) for any facet as a function of angle between the facets normal and build direction. They also prove that WCP exists for any facet if the angle is not perpendicular and can support itself by thickening the wall. Finally, the adaptive wall thickness is measured as a function of WCP. This methodology maximizes the accuracy of the fabricated self-supporting shell while minimizing the building time and material uses.
Machined sharp edge restoration for triangle mesh workpiece models derived from grid-based machining simulation
Published in Computer-Aided Design and Applications, 2018
Ziqi Wang, Jack Szu-Shen Chen, Jimin Joy, Hsi-Yung Feng
With the chamfered edge and corner triangle extraction completed, the edges and corners can be recovered. Recovering of such sharp features is often done based on some mesh de-noising methods [8],[27],[29]. However, most of the de-noising filters assume that the input data contains some noise, which obviously is not the case for machining simulation since every mesh vertex in the workpiece model is an exact tool and workpiece intersection point. An efficient numerical optimization method by Attene et al. [5] is employed in this work to split the chamfered edge and corner triangles to sharpen the edges/corners using the normals of the involved vertices. It is, thus, necessary to compute a reliable normal estimate at each involved vertex. The correct vertex normal is the normal that represents the surface normal of the adjacent non-edge feature and estimated from the corresponding adjacent triangles of the vertex.
Combined manual and automatic landmark detection for enhanced surface registration of anatomical structures: an extensive parameter study for femur and clavicle
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2020
Sanne Vancleef, Yannick Carette, Hans Vanhove, Joost R. Duflou, Ilse Jonkers, Jos Vander Sloten
Second, elasticity modulated registration is applied, consisting of a two-step iterative process. In the first step, corresponding points are identified either by manually entering the vertex indices of corresponding points of source and target mesh or by automatically identifying the corresponding points. Apart from the first iteration, this automatic corresponding point search can be skipped during following iterations. In this case, corresponding points from the previous iteration will be used. Automatic corresponding point detection is based on ray tracing, in which a line along each source’s vertex normal is drawn. The vertex normal is computed using the normal of neighbouring faces (Max 1999). A possible corresponding point is found at the intersection of this line with the target mesh. Next, the dot product of the normalised vertex normal and the normal of the face containing the intersection point are calculated. If the dot product exceeds a predefined threshold, for the dot product being between 0 and 1, these points are considered to be valid corresponding points. In case of multiple valid correspondences with the target, the intersection point with the smallest Euclidean distance is considered. In case of manually identifying corresponding points, an automatic corresponding point search is still conducted to achieve global mesh registration for areas without nearby landmarks. However, the automatically identified corresponding points for the manual landmarks are replaced by the manually entered vertex indices, and the corresponding points for the vertices surrounding the manually indicated landmarks are cleared. These corresponding points then become input for the next step in the registration framework. In the second step, the iterative N-ICP-T algorithm, described by Amberg et al. (2007), is used to control the mapping from source to target. The N-ICP-T algorithm consists of two loops: the inner and outer loop. In the outer loop, the stiffness value decreases. This decrease is defined by formula (1) and is dependent on the number of iterations of the outer loop and the start and end stiffness: