近年來隨著數位傳播技術不斷地進步，藉由逆向工程可讓生活周遭物品數位化，再配合三維列印機即可重現出實體，這之中的媒介就是利用三維模型。然而重建三維模型不管是使用輪廓法配合單一校正板或立體視覺法，所能重建的物品存在一定的大小限制和困難程度。本研究之目的即為開發一套多重二維相機校正流程，利用輪廓法技術來重建大型物體之三維模型。 本研究基於二維相機校正為基礎，設計了五種不同的校正板及對應的解碼流程，環繞物體擺放，用於辨識不同校正板與相鄰關係的擺放位置。配合平面投影轉換的估計，求得校正板空間的幾何關係，再將所有的影像校正板整合為一個全域座標系，最後配合相機校正流程求得各影像之內外部參數。本研究利用C#程式語言發展一套配合上述流程的人機介面程式，使用者僅需輸入相關資訊即可獲得建模資訊，從中觀察各個流程是否按照程序進行。 最後本研究使用三個小型物件及四個大型物件進行三維模型重建，用以驗證本研究所提出的方法之可行性。 ;With the continuous progress of digital communication technology, making object digitized and use three dimensional printers to reproduce could be achieved by reverse engineering technology. These process utilize digital three dimension model to achieve. However, whether using the Shape-From-Silhouettes (SFS) method with single 2D calibration pattern or Stereoscopic to reconstruct a three-dimension model, there are size limits and difficulty for reconstruction items. The purpose of the study is developing a camera calibration process with multiple 2D calibration patterns and a three dimension model with large-scale could be reconstructed by SFS method. In this study, five different calibration pattern and the corresponding decoding process were designed based on the two-dimension camera calibration, so the placement of patterns around objects could be identified. With the homography estimation method, the solid geometry of calibration patterns was obtained. All patterns of all images were integrated into a global coordinate system. Then both of external and internal parameters in each image were determined. The process mentioned above was integrated a human-machine interface program based on C# programing language. User can obtain the modeling information only by inputting the relevant information. Finally, this study chose three small-scale objects and four large-scale objects to demonstrate the reconstruction of three-dimension model for verifying the feasibility of the proposed method in this study.
