現今汽車鈑件之逆向工程中,廣泛應用電腦輔助設計與製造等相關技術與軟體,其中鈑件之曲面CAD模型製作,必須透過三角網格點重建曲面並調整曲面間連續性,處理流程相當耗時,倘若利用半自動化程序建立鈑件模型,將可縮短工作時程,對業界帶來莫大效益。 基於精確度與平順性之要求,於自動化曲面重建流程中,必須劃分屬性相異之區域並分別重建曲面,雖然網格區域之自動劃分研究已趨成熟,然而曲面重建結果與點資料間之誤差無法滿足業界的要求。因此,本研究主要目的即為縮小誤差,以透過區域成長等特徵區域萃取法而產生之汽車鈑件網格塊作為本研究的輸入端,依相關準則操作第二次分割與合併,針對特定網格塊予以細線化,以及利用網格點之微分幾何性質進行網格塊的邊界整形,透過上述前處理,配合對應之曲面重建流程,將誤差精度控制於業界可接受的範圍內。
Nowadays in the reverse engineering of automobile sheet metals, computer-aided design(CAD) and computer-aided manufacture(CAM) related technology and software are widely applied. The CAD model manufacturing of sheet metals, a part of reverse engineering, must reconstruct the surface through triangular mesh and adjust the continuity between the surfaces. This process is time-consuming, thus decreasing the working time by using semi-automatic process will benefits the industry. To fulfill accuracy and smoothness, mesh must be divided into different patches and for each patch, the surface is reconstructed individually. The research of automatic dividing patches is well-developed; however, the errors between surfaces and mesh do not respond to the industry expectation. Therefore, the main purpose of this paper is to achieve the industry expectation by decreasing the errors. The input data of the proposed scheme is the patches generated by region growing. The scheme is composed of the following procedures: second division and consolidation by rules, thinning specific patches, and patch's boundary rectification by differential properties of mesh vertex. By the processes mentioned above and the surface reconstruction proposed in this paper, an acceptable range is reached.