近幾年來由於科技的快速發展,有許多的產業試圖將傳統的二維平面資訊轉換為以三維立體的方式呈現,其中如3D電影阿凡達的成功,更是宣告了電影產業的3D革命,這些成功的應用都顯示了3D數位內容的潛力。 在3D數位內容的創作中,3D模型的建立是必要的一環,然而由於3D建模的高進入障礙使得目前只有專業人士才有能力進行3D建模的工作。3D建模之所以不易,主要是傳統的建模軟體必需透過平面的操作介面來操作三度空間中的物體,如此不但不直覺且在操作上也不方便。為了降低3D建模的門檻,在本研究中,我們發展了一個以擴增實境為基礎的互動式3D建模系統,此系統運用SIFT特徵及比對技術,搭配攝影機自我校正技術重建場景中特徵點的三度空間位置,再以此為基礎,利用一個有效的Perspective-n-Point演算法即時追蹤攝影機的位置與方向。利用此擴增實境技術,我們可以讓使用者在一個虛實相間的環境中對物件進行調整,進而達到3D建模的目的。
In 3D digital content creation, 3D model is an essential part. However, because of the high entry barrier of 3D modeling, only the professionals have the ability to perform the task of 3D modeling. The reason why 3D modeling is difficult comes from the fact that the traditional modeling software such as 3ds Max or Maya often modeling the objects from 2D views which is indirect and inconvenient for the operation of modeling in 3D space. In order to reduce the barrier of 3D modeling, in this thesis we develop an interactive 3D modeling system based on the technology of augmented reality. This system employs Scale Invariant Feature Transformation feature detection and matching technique as well as a camera self-calibration technology to reconstruct a set of 3D points on the scene. These 3D points as well as their projections on the image plane are then feed to an efficient camera tracking algorithm, the EPnP, to estimate camera orientation and position. Meanwhile, a primitive-based 3D modeling system is also developed. With this system, user can insert a virtual object to the real scene, then edit it via an interactive manner and finally create a 3D model for the real object using this AR platform.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。