空載多光譜影像雖具備多達11波段的良好光譜解析力和3公尺?10公尺的空間解析力，但拍攝影像時，由於飛機飛行的不穩定、高達89.52∘的視野和全景畸變使得影像幾何校正問題變得複雜且困難，同時直接影響到影像後續的應用。 本研究利用一數學模式—有理函數模式對空載多光譜影像做一快速且簡便又不失精度的幾何校正。此模式需選取大量控制點進行校正，如由人工選取如此大量的控制點，費時費力，故在本研究中使用影像特徵點選取與像元匹配方式，自動選取大量且均勻之控制點，隨後在控制點中加入高程資訊，利用有理函數模式進行影像的幾何校正。 研究成果顯示於各種高程起伏情況之小範圍影像，使用RFM高階模式進行幾何校正，可達近2個像元之精度，如果選用高階以上的校正模式對大範圍影像進行處理，亦能有3像元內的精度產生。所以在不管高程起伏情況下，使用小影像進行RFM高階模式幾何校正較為理想；若欲使用大影像進行RFM高階模式幾何校正，此時校正精度視地形起伏大小而定，大約在3像元左右。證實本研究方法可提供空載多光譜影像一種便捷且自動化的幾何校正方式。 Although Airborne MSS imagery have spectral resolution owning eleven bands at most and spatial resolution within three to ten meter , the geometric correction of an imagery become complex and hard , in terms of the unstable aerial sailing , up to 89.52° of FOV and panoramic distortion. The research provides the approach used to deal with the Airborne MSS imagery’s geometric correction with a mathematical model—Rational Function Model (RFM) that can correct imagery simply , quickly ,and accurately . The RFM needs numbers of control points for geometric correction . Selecting numbers of control points is time-comsuming and labored . In the research , imagery feature selecting method and pixel matching method pick up numerous and uniform control points that contain DTM information automatically, then imagery geometric correction set out to use RFM. The research results explain that any terrain relief of a small range imagery using RFM’s high order model for geometric correct could reach 2 pixel accuracy ,and It is within 3 pixel accuracy in the large range imagery . If in spite of terrain relief , a small range imagery using RFM’s high order model for geometric correction is good , and it depends on terrain relief in a large rage imagery. The accuracy is about 3 pixel.The results demonstrate that the research approach provide a fast and automatic geometric correct method for Airborne MSS imagery.
