以Pix4Dmapper進行UAV影像之空間資料產製

THE USE OF INNOVATIVE SOFTWARE PIX4DMAPPER TO OPTIMIZE THE PROCESS OF GENERATING SPATIAL DATA FROM UAV'S AERIAL IMAGES

徐金煌(Robert Hsu) ； 張國楨(K. C. Chang)

《中國土木水利工程學刊》 27卷3期 (2015/09) Pp. 241-246

https://doi.org/10.6652/JoCICHE.201509_27(3).0009

摘要

無人飛行載具UAV(Unmanned Aerial Vehicle)近年來發展快速，加上自動導航技術的提升與穩定的航拍取像設備，使得UAV可以更安全、更自動化的方式取得高解析度的航拍影像，此外UAV亦能執行低空雲下自動化取像，相當適合作為防救災應用。平時可進行長期的環境變異資料監測；災時可快速取得影像資料，了解災害現場影響等。但傳統使用UAV進行正射影像產製，必須透過地面控制測量、空中三角測量與立體製圖等方式，才可以取得正射影像與地形資訊，無法滿足災害發生時需快速取得空間資料之需求，因此如何透過UAV所拍攝的影像資訊快速產製正射影像與地形資訊為本研究之重點。本研究將以Pix4Dmapper 為先進技術軟體，探討以UAV航拍取像後，透過先進技術取得正射影像與地形資料之可行性。結果顯示，UAV拍攝所得影像透過Pix4Dmapper進行影像匹配三維點生成地表模型，快速生產正射影像，並可以不需經由人工介入的空中三角測量步驟回推內、外方位參數。並以苗栗後龍溪測區為例，在涵蓋範圍4 km^2，影像數量354張，在自動化的作業流程下，可在7小時內取得相關空間資料，經檢核後，正射影像之平面精度可達15 cm。

關鍵字

無人飛行載具 ； Pix4Dmapper ； 影像匹配

並列摘要

An Unmanned Aerial Vehicle (UAV) industry is rapidly expanding for the past few years. The enhancement of an automatic system technology and the stable image capture equipment provides an ability for UAVs to adapt changing condition automatically and capture stable quality of high resolution ground images. The UAVs are also able to capture ground images at a low altitude (about 500 to 700 meters height) which could provide advanced use of disaster management, including disaster area monitoring, information gathering, and rapid response to disaster. However, a process of producing orthophotos and a digital surface model (DSM) from the UAVs' images in traditional methods, which need to set three control components first, including aerial triangulation, ground control points, and stereo photogrammetry, is time-consuming. It cannot react to the rapid demand of disaster response. The study focus on optimizing the process and on reducing time-consumption to acquiring orthophotos and the DSM from UAVs' aerial image by using Pix4Dmapper. The Pix4Dmapper could automatically process orthophotos and the DSM purely using aerial image captured by UAVs. The Houlong River, Miaoli, Taiwan case is used in the study. The Houlong River covers an area about 4 sq. km, and the analysis of aerial images of the River are 354 images. The automatic process workflow of acquiring orthophotos and the DSM from the aerial images by the Pix4Dmapper is within 7 hours. And the planer accuracy of otrhophoto achieve 15 cm.

並列關鍵字

UAV ； Pix4Dmapper ； image matching