透過您的圖書館登入
IP:18.118.145.114
  • 期刊

高酬載無人機搭載五相機攝影系統之影像拼接及空三平差精度分析

IMAGE MOSAICKING AND AERIAL TRIANGULATION ACCURAY ANALYSIS FROM FIVE CAMERA IMAGING SYSTEM EQUIPPED BY HIGH PAYLOAD FIXED-WING UAS

摘要


無人機具高機動性與低人員操作風險之優點,但卻限於載重、容量與續航力,僅能搭載單台小像幅相機進行小範圍的調查。本研究以高酬載定翼無人機搭載五台小像幅相機組成的多相機攝影系統,透過傾斜同步攝影進行多影像拼接,以擴大拍攝範圍及減少航帶數。本研究之多影像拼接法,首先透過相機率定求得透視投影轉換的各項係數,將原始影像糾正與拼接。接著分析糾正影像重疊區的套合誤差,並得到尺度與平移兩系統性誤差,經逐一改正後證實拼接影像可達次像元精度。最後由空三平差製圖程序分析拼接影像之絕對定位精度仍能符合1/1000製圖規範之要求。

並列摘要


Unmanned aerial system (UAS) has advantage of higher mobility and lower risk comparing with manned airplane. However, most UAS were restricted by its low payload, small capacity and short endurance time thus can only carry one single small format camera for small area investigation. In this study, a five small format DSLR cameras composed imaging system is proposed and mounted on a high payload UAS for large coverage image collection. In which, one nadir and four oblique views imagery were acquired at the same time and mosaicked into larger format image to enlarge ground coverage, increase the imaging geometric strength and overall positioning accuracy. The multi-image mosaicking algorithm is first derived the projective transformation coefficients and interior orientation parameters from indoor camera calibration. To analyze the co-registration residuals between the overlapped areas of two neighbored images, two systematic errors, i.e. scale and displacement, are obtained and applied to refine the image mosaicking results. Accuracy analysis was performed by conventional aerial triangulation procedure. Result shows that the mosaicked image can achieve sub-pixel accuracy. Meanwhile, its absolute positioning accuracy can also meet the 1/1000 topographic mapping standard.

延伸閱讀