以空載光達同步航空攝影製作高程模型品質探討

Abstract

空載光達可快速獲得大量點雲，為高程模型製作之一項重要作業方式。著眼於同時提供正射影像產品，目前空載光達各配置中幅航空攝影相機同步作業。雖然航空影像之攝取並非為數值高程模型之製作，但是仍有此一潛力。本研究以大安溪蘭勢大橋附近河道測量為例，以空載光達成果為檢核資料，探討伴隨空載光達所得之中像幅數位航空影像，採用自動匹配方式製作數值高程模型之品質，並探討數位相機受輻射畸變差對高程誤差的影響。 研究成果顯示，於明確定義點誤差分析，高程較差RMSE為0.5528公尺。於坡度與誤差分析，誤差隨坡度愈大則有愈大的趨勢，10°坡度間隔之高程較差RMSE介於1.4685~7.9421公尺之間。於植被覆蓋與誤差分析，人工林之高程較差RMSE可達1.0046公尺，自然林則可達1.4355公尺。 中像幅數位量測相機受透鏡輻射畸變差影響相當大，比較進行透鏡輻射畸變差改正與未改正成果，平均改正量為11.95公尺，改正量標準偏差為16.16公尺。以空載光達點雲成果為檢核資料，進行改正之RMSE為3.56公尺，不進行改正則為19.80公尺。

Airborne LiDAR contribute a large number of 3D point clouds, which is one of the main methods to derive DEM. However, the aerial image taken simultaneously with an airborne lidar is potential for producing DEM. Focusing on provide orthoimage simultaneously, airborne LiDAR equipped with digital middle format frame camera to obtain images, which is potential for producing DEM. In consequence, comparing with the airborne lidar, this study using riverway survey data of Taan River Lan-Shi bridge area analyses the quality of DEM that using automated matching aerial images, as well as the influence of height accuracy caused by radial distortion. It is shown that the RMSE of the height difference value is 0.5528 m in the Well -defined point anaysis, while it is 1.4685~7.9421 m in 10° slope in the slope and error analysis. Steeper slope with the higher error is also proved. In vegetation coverage and error analysis, the RMSE of the height difference value in plantations and natural forest is 1.0046m, 1.4355 respectively. The influence caused by the radial distortion was demonstrated. Comparing the heights resulting from automated matching with the corrected and un-corrected images,the bias is -11.95m, and the standard deviation is 16.16m. Comparing with the airborne lidar, the RMSE from the corrected image is 3.56m, while 19.80m for the un-corrected.