Abstract: A sem-i automatic algorithm based on active window line segment matching is proposed for extracting road centerlines from high spatial resolution remote sensing images.Road centerline is tracked automatically starting from a user-specified point by defining a series of active template windows,and followed by using thresholding,line segment matching and an improved SSDA.In addition,some kinds of flexible user interventions are allowed in our algorithm in case automatic tracking fails.Experiments on0.61m-resolution QuickBird and1m-resolution IKONOS satellite images show that our proposed algorithm is capable of rapidly and accurately extracting main road centerlines in addition to its good robustness to noise.