旋翼UAV影像製作千分之一都市區局部區域正射影像之探討

An Investigation into the Generation of 1/1000 Orthoimages for Local Urban Area by VTOL UAV Images

正射影像（Orthoimage）是將航拍影像上的像點高差和影像傾斜位移移除，可提供正確的幾何和輻射資訊作地理資訊的底圖、或套疊數值高程模型（Digital Elevation Model, DEM）作為三維景觀模擬供相關研究之用。以往千分之一正射影像製作仰賴大型航空攝影飛機酬載量測型相機進行航空攝影測量取像，並依此進行資料正射影像製作。近年來無人飛行載具（Unmanned Aerial Vehicle, UAV）發展快速，以其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、甚至可在低空雲下飛行取得高解析度的航拍影像，相當適合作為局部區域製作正射影像的航拍取像系統，尤其是旋翼型UAV雖航程較小、抗風性較定翼機差，但因可垂直起降、機動更強，更適合用於都市地區局部製作正射影像之航拍取像。然而旋翼型UAV因酬載能力僅能酬載非量測型相機航拍取像，且因其像幅涵蓋範圍有限，因此取得之UAV影像是否能局部製作符合都市地區千分之一精度之正射影像將是本研究探討之重點。實驗結果顯示於航線規劃後以旋翼型UAV航拍取得品質佳之影像，並透過自率光束法空中三角測量平差，解決以近距離近景攝影測量程序率定之相機參數無法完整描述於空中航拍取像時的相機參數問題之後，再限制每張影像製作正射影像之範圍，證實以旋翼型UAV影像能製作都市區局部區域符合千分之一精度要求之正射影像，而本研究最後並依據實驗結果提出結論與建議。 The image displacements on aerial images due to photographic tilt and topographic relief should be exactly removed to generate orthoimages. Thus orthoimages can provide the correct geometrical and radiometric information for basic image maps for geographic information, or they can overlap with digital elevation model for landscape simulation for advanced studies. In the past, 1/1000 orthoimage generation relies on large-scale aerial aircraft with metric cameras for collecting the aerial images. In recent years, the development of UAVs (Unmanned Aerial Vehicles) is quite fast. Due to the automatic navigation technology, it makes UAVs be a stable platform to collect high-resolution aerial images much more safely and automatically. Also, UAVs can acquire the aerial images with low-altitude and under cloud, UAVs become quite suitable aerial image collection system to locally produce the orthoimages. Particularly, VTOL (Vertical Take Off and Landing) UAVs can take off and land vertically, therefore they are more flexible and more suitable for acquiring aerial images in urban areas to generate 1/1000 orthoimages locally. The payload capacity of VTOL UAV is much more limited than the fixed-wing UAV, thus only the non-metric camera can be implemented to collect aerial images. Additionally, the images are often limited ground coverage. Therefore, this study will focus on the investigation into the generation of 1/1000 orthoimages locally by using these collected aerial images. Based on the experimental results, after planning and acquiring the aerial images with good quality, self-calibration bundle adjustment for aerial triangulation is performed to overcome the imperfect camera parameters calibrated by the procedure of close-range photogrammetry. Next, by limiting the used image content it proves that the aerial images collected by VTOL UAVs can be used to generate the orthoimages locally in urban area. Finally, the suggestion will be presented based on the experimental results.

The image displacements on aerial images due to photographic tilt and topographic relief should be exactly removed to generate orthoimages. Thus orthoimages can provide the correct geometrical and radiometric information for basic image maps for geographic information, or they can overlap with digital elevation model for landscape simulation for advanced studies. In the past, 1/1000 orthoimage generation relies on large-scale aerial aircraft with metric cameras for collecting the aerial images. In recent years, the development of UAVs (Unmanned Aerial Vehicles) is quite fast. Due to the automatic navigation technology, it makes UAVs be a stable platform to collect high-resolution aerial images much more safely and automatically. Also, UAVs can acquire the aerial images with low-altitude and under cloud, UAVs become quite suitable aerial image collection system to locally produce the orthoimages. Particularly, VTOL (Vertical Take Off and Landing) UAVs can take off and land vertically, therefore they are more flexible and more suitable for acquiring aerial images in urban areas to generate 1/1000 orthoimages locally. The payload capacity of VTOL UAV is much more limited than the fixed-wing UAV, thus only the non-metric camera can be implemented to collect aerial images. Additionally, the images are often limited ground coverage. Therefore, this study will focus on the investigation into the generation of 1/1000 orthoimages locally by using these collected aerial images. Based on the experimental results, after planning and acquiring the aerial images with good quality, self-calibration bundle adjustment for aerial triangulation is performed to overcome the imperfect camera parameters calibrated by the procedure of close-range photogrammetry. Next, by limiting the used image content it proves that the aerial images collected by VTOL UAVs can be used to generate the orthoimages locally in urban area. Finally, the suggestion will be presented based on the experimental results.