Semiautomatic Airport Runway Extraction Using a Line-Finder-Aided Level Set Evolution

In recent years, airport runway extraction has become increasingly important for various engineering applications. Existing approaches for airport runway extraction primarily focus on locating the airport roughly, i.e., determining whether an airport is present or not, but not delineating the airport runway accurately. This study develops a novel method for semiautomatic airport runway extraction from Google earth images by integrating a long straight line finder and a region-based level set evolution (LSE). Specifically, we start by detecting the long straight lines that most likely represent airport runway boundaries in the original images. Then, based on the extracted lines, we propose a method for semiautomatic generation of initial level curves for the LSE. Furthermore, for accurate extraction of the entire airport runways, a fast region-based LSE is used to evolve the initial level curves toward the desired boundaries. Experiments validate that the proposed method is capable of semiautomatically extracting objects with complex geometrical shapes and topological structures from challenging backgrounds. Compared with other state-of-the-art approaches, the proposed method has much fewer parameters and is more computationally efficient while achieving object extraction accuracy comparable to other approaches.