以數張影像迅速建立一物體之3D模型

Abstract

在本論文中，我們提出了一個既簡便又快速的方法來重建物體的3D模型。由於我們使用了目前最為流行的數位相機來拍攝物體的六視圖影像，因此整體的實驗設備相當便宜且實驗環境也相當的簡單而容易設立。我們提出了一些演算法來處理這些影像，包括背景濾除和各視圖對應等，並進一步地將所有部分串連起來成為一個完整的流程來進行整個物體之3D模型的重建工作。經由四個物體實際重建後的結果顯示，我們所提出的方法確實是相當地迅速，且重建後所展示的模型也十分地逼真，結果顯示所提方法極具實用性。

In this thesis, we propose a method to quickly reconstruct the 3D model of an object in a simple and easy way. Because we use the most popular digital camera at present to capture the six-view images of the object, all experimental equipments are quite inexpensive and the experimental environment can be easily set up. We propose some algorithms, such as background-filtering and view-corresponding, to deal with these images, and then link all algorithms to become a complete procedure to reconstruct the 3D model of the object. After reconstructing four objects, results show that the proposed method is indeed quick and the 3D model that has been reconstructed is almost identical to the real object itself. Experimental results show the applicability of the proposed method.