Abstract: The traditional airport subgrade flatness detection has the problems of long detection time, large errors and one-sidedness of detection results, which make it difficult to get an effective evaluation. In order to realize the rapid, comprehensive and accurate detection of airport subgrade evenness, this paper proposes a method to detect subgrade evenness by using UAV scanning subgrade image information, obtaining a three-dimensional real model through real scene modeling and calculating international evenness index (IRI). Taking the foundation construction of an airport taxiway as an example, the flatness detection results of vehicle-mounted laser flatness meter and UAV real scene modeling are compared. The results show that it is feasible to use UAV real scene modeling for flatness detection, and this method can realize the flatness detection of any full-length road section quickly and accurately, and it is no longer limited to the flatness detection on the straight line of the driving lane, avoiding one-sidedness and contingency.