- 學位論文
語意參數設計於自由幾何產品之客製化
Semantic Parametric Design of Freeform Product Customization
摘要
3D模型被普遍運用於設計、工程製造等多種領域，現今的電腦輔助設計系統大多可幫助設計者從3D模型自動產生2D的工程繪圖，但若要要建立複雜的3D模型仍需花費許多技術與時間。除了一般的3D工程模型之外，爲了符合顧客的需求，具有自由幾何特性(freeform)之曲面也愈來愈常被使用於產品客製化上。然而，自由幾何產品之建模通常會比一般工程產品複雜，所需要耗費的時間成本也就更多。本研究以此問題為出發點，期望將以往僅被運用於工程產品上之參數設計(Parametric design)方法，應用於自由幾何之產品建模，幫助設計者建立或修改模型，增進在設計研發階段之效率，達到降低時間及金錢成本之目的。 參數設計是一種利用物件幾何限制來修改或產生模型幾何的方式，但對於自由幾何物件，我們卻難以明確定義其幾何作為參數限制，因此，參數設計以往並不常被使用於自由幾何物件之建模，學界亦缺乏關於此方面之探討。本研究之首要目的，即是針對自由幾何物件發展一套語意參數設計(Semantic Parametric Design)演算法，使設計者可直接透過語意參數的修改，產生符合顧客需求之自由幾何模型，以達到客製化之目的。此演算法以統計方法中的線性迴歸作為基礎，必須先建立數量足夠的模型作為參考模型組，每個參考模型皆有對應的語意設計參數，在符合系統限制之下，本研究使用線性迴歸以求得模型三角網格(Triangular mesh)與語意設計參數之對應關係。基於此演算法得到之網格與參數對應關係，設計者可透過直接修改參數來得到客製化之模型。本研究發展之方法具有高度實用性，在服飾設計方面，可用於衣服、珠寶等自由幾何產品；在生物應用方面，可修改人體模型，建立人體資料庫。此外，以MATLAB為環境之演算法亦具有高度的系統合適性。
並列摘要
A 3D solid model is used in various processes such as design, engineering evaluation, drafting, manufacturing, and so on. Most commercial CAD tools support functions to generate engineering drawings from solid models automatically. However, It is a time-consuming and skill-required job to create a complicated 3D solid model. Parametric modeling is an approach to construct and to modify model geometry with geometric constraints. Nowadays, freeform surfaces are frequently used in a product design in order to meet customers’ aesthetic needs. Freeform modeling requires considerably more time and effort than ordinary mechanical modeling. It is difficult to impose parametric constraints on freeform models since the shape of freeform models is fully arbitrary as its name implies. As a result, parametric modeling has not been popular in freeform modeling. The primary objective of this research is to simplify the geometric processing required by freeform product customization. We develop a statistical model and a reliable Semantic Parametric Design algorithm applicable to various areas including human body shape, and garment design. The algorithm helps user construct a new model which is conformed to its semantic parameter and the user’s request. We build a reliable object database and compute models and semantic parameters as a statistical linear system. After linking each parameters and triangular mesh of corresponding models, we can generate a new model fitted with the user’s needs by this association. Besides, we provide a method to inspect if the new modified value is reasonable. This makes great helpful for design efficiency.