近年來使用數位視訊攝影機(digital video camera)來進行環境監控越見普及，但此監測方式大部份仍局限於傳統的地面調查。因此，本研究整合遙控飛機和視訊攝影機，進行空中之監控及調查。在使用及應用上，通常會有某些特定的用途，如較大涵蓋範圍之監控，需要把視訊影像以照片方式輸出並進行鑲嵌處理，以提供一個完整而連續的研究區域影像，藉此利於判釋及展示。本研究提出一套完整的鑲嵌圖製作流程，其主要工作為利用自動化萃取及匹配特徵點的方式進行處理。一般而言，視訊攝影機有高取樣頻率的特點，因此在使用上會產生大量連續且高重疊度之立體影像，藉重疊區域內高立體重疊度之特性，可獲得充份的影像控制點以供立體對匹配使用。本研究即以此特性研發鑲嵌圖自動製作的流程，主要工作是自動化特徵點的萃取及匹配，並建立連續相鄰影像間的映對(mapping) 關係，然後經適合的坐標轉換模型，轉換至參考坐標系中，並接合各相鄰影像完成鑲嵌圖之製作。測試的成果，顯示所提出的方法，確實能夠提供一個高效率、自動化、低成本的空載視訊影像鑲嵌圖之製作。 This paper presents an experimental method for mosaicking aerial video image. Recently, it has been a growing interest to use digital video camera for environmental monitoring. However, the use has been mostly limited to the ground investigation. This study integrates the video camera with the unmanned vehicle to become an aerial monitoring and mapping system. In practical use, it has been always some need to transform the aerial video image into a single photo. This will provide a useful visual interpretation for the area covered by the aerial video image. In classical photogrammetry, the overlap area between successive photos fundamentally provides the basis for photo mosaic. In general, the video image has the capability to generate numerous successive images with high degree of stereoscopic overlap area. Accordingly, there are sufficient control points for stereoscopic matching. The proposed method develops an image matching procedure to extract feature points from overlap areas. Then the image mosaic can be effectively performed using these feature points. The test results indicate that the proposed method is able to provide an efficient and low-cost system for mosaicking aerial video image.
