裂縫的尺寸是決定混凝土橋梁壽命的關鍵要素之一。當前，傳統橋梁檢測仰 賴檢測員利用梯子或纜繩攀爬橋梁進行作業，不僅相當耗時且費力，此方法 的安全性也有些疑慮。為改善此問題，近期有許多研究自製機械手臂加載相 機並透過影像處理偵測及量測裂縫。然而，機械手臂所費不貲。因此，本研 究設計並實驗一套運用無人機承載高解析數位相機檢測橋梁裂縫的方法。 具體而言，本研究透過空間三角平差將涵蓋地面控制點的遠拍影像及具備 清楚裂縫特徵的近拍影像透過影像間的共軛點將所有影像套合在一個絕對 坐標系統。其中，本研究利用三種方法產製共軛點:(1)人工點選橋梁現 有特徵作為共軛點。(2)隨機挑選由 Scale-Invariant Feature Transform(SIFT) 產生的共軛點。(3)隨機挑選由 SIFT 產生的起始點再以 ERDAS Leica PhotogrammetrySuite(LPS)提供之匹配流程(AutomaticTieGeneration)產製 共軛點。在確立絕對坐標系統後，所有的數位影像將被重新採樣為每像元 0.1 毫米的正射化影像來進行後續的裂縫量測，並運用人工量測及反曲點辨 識裂縫邊界的兩種方法進行量測比對。本研究在桃園市楊梅區愛鶴橋進行 現地實驗。與測驗員的量測數據相比，以人工產製共軛點並量測裂縫的方展 現最佳成果，可達到 0.18 毫米的均方根誤差、0.15 毫米的平均誤差、25.41百分比的相對誤差。而結合 SIFT 和 LPS 自動化產製共軛點及人工量測裂縫 的方法達到 0.46 毫米的均方根誤差、0.4 毫米的平均誤差、72.42 百分比的 相對誤差。;Crack development is a clear indicator to the durability of concrete bridges. Traditional bridge inspections which rely inspectors to climb on bridges with lift cars are not only unsafe for inspectors, but also time and labor consuming. While there have been some researches designing robotic arms carrying digital cameras and applying image processing for crack detection, robotic arms are expensive. Therefore, this research proposes a solution that applies unmanned aerial vehicles (UAV) and high-resolution digital cameras to measure concrete bridge cracks. To be specific, two types of images are taken, which are close-up images that can observe cracks more clearly and long-range images that covers ground control points. Afterward, we register these two types of images to establish the absolute coordinate system with ground control points and tie points through the block triangulation. This research examines three approaches of generating tie points: (1) Manually select tie points with features on the bridge such as nails and dots. (2) Randomly input tie points generated from Scale- Invariant Feature Transform (SIFT). (3) Randomly input tie points generated from SIFT as initial tie points and perform automatic tie generation with Erdas Leica Photogrammetry Suite (LPS) image matching module (Automatic Tie Generation). Afterward, close-up images are processed into ortho-rectified images with 0.1mm pixel size for crack size measurement. Crack sizes are finally determined by a manual measurement approach and an inflection point approach for comparison. An experiment was conducted on Ai He concrete bridge located in Yangmei District, Taoyuan City, Taiwan. By comparing to in-situ measurements from three surveyors, the proposed solution that manually selects tie points and measure crack sizes produces the best result with 0.18 mm root mean square difference (RMSD), 0.15 mm mean difference (MD), and 25.41% mean relative difference (MRD). Furthermore, manual crack size measurement with automatic tie point generation with SIFT and LPS achieve 0.46 mm RMSD, 0.4 mm MD, and 72.42% MRD.
