整合三維房屋模型與二維影像之套合、匹配、及牆面重建之研究

Abstract

三維房屋模型是數碼城市及適地性資訊服務中重要的數位內容。本計畫使用都市等 級三維房屋模型(OGC LOD2)及二維近景影像建立建築等級三維細緻房屋模型(OGC LOD3)。研究重點為三維房屋模型與二維影像之空間套合、特徵匹配及牆面幾何重建。 第一項工作目標為三維房屋模型與多重疊影像之特徵匹配，研究重點為多重疊影像之匹 配技術，研究中使用三維房屋模型做為輔助資訊，結合輻射約制、參數約制及幾何約制 進行線形特徵之匹配，最後以空間前方交會完成三維牆面結構線之萃取；第二項工作目 標為半自動化三維細緻房屋模型之牆面幾何重建，研究重點為牆面幾何重建紋理敷貼， 研究中使用半自動化方式選取初始模型，經模型庫之自動化模型擬合產生牆面幾何模 型，完成建築等級三維細緻房屋模型重建。

Three-dimensional building model is an important digital content for Cyber City and Location-based Service. This project plans to establish an OGC-LOD3 building model from OGC-LOD2 building model and close range images. The core technology of this research contains space registration, feature matching, and facade reconstruction by the integration of OGC-LOD2 building models and close range images. The works in the first item will include the extraction of 3-D facade structural lines using 3-D model and 2-D images. Linear feature extraction and matching will be the main focus. Feature matching will utilize radiometric, geometric and parameter constrains. The major components will include feature extraction, image matching, and space intersection. The works in the second item will deal with the reconstruction of 3-D facade by the integration of 3-D model and 2-D images. Semi-automatic model fitting and 3-D facade texture mapping will be the major components.