三維網格模型的特徵辨識技術不僅運用於幾何模型化(Geometric modeling)與電腦輔助設計(Computer aided design)領域中，另外包括了各種網格處理的相關技術，也都以特徵辨識為基礎，其運用的範疇非常廣泛。現有CAD/CAM軟體的特徵辨識技術對於特徵分佈較為規則的機械外形元件有較好的辨識效果，而非規則外形的造型模型或是人體組織模型，則無法得到合適的辨識效果，通常需輔以人為手動的方式定義特徵位置，造成效率上的問題。本計畫的主要目的是以網格分群(Mesh clustering)技術為演算法核心，發展一個能夠應用於非規則造型模型與骨組織模型的特徵辨識演算法，並且整合網格處理、曲線嵌合等技術，讓網格模型經由本程序能夠迅速的得到合適的特徵曲線。對於特徵辨識演算法的發展，預期達成以下目的：(1)發展非規則模型的特徵辨識技術，(2)以半自動化介面取代手動建構，(3)演算法參數之實驗與分析、增進演算法對網格雜訊之強健性。擬發展技術的應用則包括非規則造型之模型特徵擷取和骨頭斷裂面與輪廓線擷取，本研究中將以數個範例證明此一技術在這些應用的可行性。Feature extraction is not only an important tool in many fields of geometric modeling and computer aided design, but also a useful technique in many applications of mesh processing. The commercially available CAD/CAM software is generally valid for models composed of regular features, such as quadratic surfaces. For models composed of irregular shape, however, manual operation is generally required to separate each of the features one by one. The primary objective of this study is to develop a features extraction process by means of iterative clustering algorithm for the extraction of irregular features in the cloud points. It mainly consists of the following three procedures: mesh processing, features extraction and curves fitting, which will be integrated to assist the operator obtaining suitable feature curves more quickly and easily. Specific targets of the proposed project are as follows: (1) developing the feature extraction technique for models of irregular shape, (2) establishing a semi-automatic interface so that the operator can manipulate the process easily, and (3) improving the robustness of the proposed algorithm through experiments to analyze and adjust the parameters and data. The applications of the proposed method include feature extraction of irregular geometric model and bone fracture zone detection. Various examples will be presented to demonstrate the feasibility of the proposed method in these applications. 研究期間：10008 ~ 10107
