三角網格特徵線萃取與分割於汽車鈑件之應用

Application of Feature Line Extraction and Triangular Mesh Segmentation for Automobile Sheet Metals

李康豪(Kang-Hao Lee)

指導教授 ： 尤春風

國立臺灣大學/工學院/機械工程學研究所/碩士(2010年)

https://doi.org/10.6342/NTU.2010.00232

摘要

逆向工程是從現有的實物中或樣品模型中，透過三次元量測儀器取得點資料，再由點資料將曲面重建。在逆向曲面重建裡，曲面重建無法直接處理龐大的網格資料，必須將網格資料以區域來處理。本研究針對大量三角網格資料提出一套處理網格分割與分群之程序，不僅可快速有效的縮短曲面重建時程，也提升曲面與原始網格的精確度。
網格分割與分群程序中所用的演算法包含三個部分：特徵線萃取、網格面分割以及網格分群。首先，特徵線萃取是以曲率變化及透過門檻值的過濾產生特徵點，再將所產生的點連接成為特徵線，並進行特徵線延伸取得封閉網格塊，同時利用網格拓樸關聯性進行三角網格面分割，最後將封閉網格塊進行網格分群，取得完整區塊。

關鍵字

逆向工程 ； 三角網格 ； 特徵線萃取 ； 網格面分割 ； 網格分區域

並列摘要

Reverse engineering is a process of reconstructing CAD models based on an existing object. With the help of a coordinate measuring machine (CMM), we can model the data points into triangular meshes, then reconstruct the surface models from the triangular meshes. In surface reconstruction, it is impossible to construct a surface with a huge number of data points; therefore, we must segment the meshes into smaller regions. In this paper, a triangular mesh segmentation procedure is proposed. The procedure not only efficiently shortens the surface reconstruction, but also improves the accuracy of the reconstructed model to the existing object.
The basic approach consists of three steps: feature line extraction, mesh segmentation, and finally mesh separation. Firstly, feature line extraction is based on the changes of principal curvatures and filter through a threshold value to generate feature vertices. Next, join the feature vertices to form feature lines and extend them to form closed loops. In the meantime, using the topological information of triangular mesh, we can start mesh segmentation. Lastly, with the help of the closed loops, we can decompose the mesh models into smaller regions.

並列關鍵字

reverse engineering ； triangular mesh ； feature line extraction ； mesh separation

參考文獻

[15] 楊偉倫，2007，三角網格錯誤偵測與特徵線萃取，國立台灣大學機械工程學研究所碩士論文。

[2] Goldfeather J, V Interrante. 2004. A novel cubic-order algorithm for approximating principal direction vectors. ACM Transactions on Graphics 23(1):45-63.

[3] Hubeli A, Gross M. 2001. Multiresolution feature extraction for unstructured meshes. Proceedings of the Conference on Visualization, 287-294.

[4] Kim HC, SM Hur, SH Lee. 2002. Segmentation of the measured point data in reverse engineering. International Journal of Advanced Manufacturing Technology 20(8):571-580.

[6] Razdan A, MS Bae. 2003. A hybrid approach to feature segmentation of triangle meshes. Computer-Aided Design 35(9):783-789.