Abstract: | 逆向工程為從實物或樣品,經過量測設備掃描得到點資料,進而重建CAD模型,達到幾何模型重建的一種技術。而透過量測設備掃描點資料、擷取重建CAD模型時所需要的尺寸,與比較點資料與CAD模型間的誤差則是屬於檢測的應用範圍。所以,完整的產品模型重建流程,除了逆向工程中重建CAD模型的相關建構技術,也必須搭配檢測方面的技術應用。在本研究中,所謂的最佳化嵌合理論,指的是如何從兩筆幾何資料中,找到最佳的對應關係,使得兩筆幾何資料能夠達到最佳的配合,而幾何資料可以是點資料、曲線資料與曲面資料。所以,最佳化嵌合理論可說在逆向工程與檢測中的應用非常的廣泛。 逆向工程的相關建構技術,其中最重要的一環則是自由曲面嵌合技術,曲面嵌合代表著透過某些曲面嵌合演算法計算出曲面的控制點,使得曲面與點資料間的誤差能夠最小化。而點資料的取得,目前大部分都是透過非接觸式量測設備掃描得到,所以就目前逆向工程中大部分的點資料均為亂點資料。所以,在本研究中發展出四種針對亂點資料進行曲面建構的技術,其中有曲面貼覆、大面嵌合、剪裁曲面嵌合與井字形邊界輪廓拘束性嵌合,最後並透過球鞋空氣袋的範例,來整合介紹上述四個演算法。 而最佳化嵌合理論於檢測上的應用,在本研究中則是著重在如何使得量測點資料與CAD模型間的誤差能夠最小化,其中量測點資料的取得又可分為使用接觸式與非接觸式量測設備兩種。接觸式量測設備指的是CMM,在檢測的過程中,CMM的座標系統常常必須與CAD模型座標系統一致,但如果工件上無法提供足夠的幾何特徵,如:平面、圓柱與球面等,就無法使用傳統的3-2-1法來建立座標系統。另外對非接觸式量測設備而言,掃描的過程中,並不需要建立座標系統,就可以得到大量的點資料,往往這些點資料的座標系統與其CAD模型座標系統都會有位置上的錯置關係,此時如果要檢視某個已定義剖面的誤差時,就必須將點資料移動到CAD模型上,亦即將點資料的座標系統移動到符合CAD模型的座標系統上,讓兩筆幾何資料間的誤差能夠最小化。所以在本研究中,針對接觸式與非接觸式量測設備,發展出兩種座標定位的方法與流程,可以找出最佳化的座標轉換矩陣,來解決上述的問題,最後並透過一些範例與測試數據,來驗證演算法的穩定性與流程上的可行性。 Reverse engineering is a technique of using the scanning machine to obtain the digitized points and reconstructing the CAD model for an existing object or prototypes. The applications of inspection include acquiring the digitized points, getting the dimensions needed for constructing the CAD model and comparing the deviations between the points and the CAD model. Therefore, a complete CAD reconstructing process of products requires the combination of reverse engineering and inspection. In this study, the best fitting algorithm means finding the optimum relation between two different geometric data, which could be points, curves and surfaces. Thus, the best fitting algorithm can extensively be applied in reverse engineering and inspection. The surface fitting is an important technique in reverse engineering. It calculates the control points of a surface by using some algorithms to minimize the errors between the points and the surface. At present, the digitization of an object is generally via optical scanning machines, which yields random point data. Hence, there are four algorithms developed in this study for different applications. They are surface wrapping, extended surface fitting, trimmed surface fitting and constrained boundary surface fitting. Several examples are presented to illustrate the feasibility of each algorithm. An air bag is finally used to demonstrate the integrated application. The application of the best fitting algorithm in inspection focuses on minimizing the deviations between the measurement points and the CAD model. The measurement points come from a contact or non-contact scanning machine. The contact scanning machine usually means the CMM. The part coordinate in the inspection should ideally be the same as that of the CAD model. However, if there are not enough features such as planes, cylinders and spheres, it can not use 3-2-1 method to build the part coordinate. For the non-contact scanning machine, it doesn’t need the setting of the part coordinate before the digitization is performed. Therefore, the coordinates of the digitized points and the CAD model are usually mis-matched. When we want to inspect the deviations of some defined sections, we must translate the coordinate of the points to that of the CAD model, and minimize the errors between them. Thus, two localization algorithms are developed in this study for contact and non-contact scanning machines. It can calculate the optimal coordinate transformation matrix to minimize the deviations between the points and the CAD model. Several examples are presented also to demonstrate the feasibility of the proposed localization algorithms. |