MULTI-DATE WIND TURBINE DETECTION ON OPTICAL SATELLITE IMAGES

Abstract. Satellite imaging is shifting from the photo-interpreter era to one of automatic monitoring. Indeed, the vast amount of data provided by the recent constellations of satellites, performing recurrent observation of every point on the globe, can only be handled by automatic methods; controlling false detections is thus crucial. The low costs of those satellites often imply lower resolution; the fusion of multi-date images can compensate to some extent the low resolution. Given their future role in the energetic transition and their spread over countries or continents, monitoring wind turbines is a natural candidate for such studies. This work details an algorithm for automatic, multi-date wind turbine detection on low resolution optical satellite images. The method is based on the a contrario statistical approach to provide a control of false detections and exploits the geometry of wind turbines’ shadows and hubs.