Remote sensing for greenhouse detection from stereo pairs of WorldView-2 satellite

Abstract

The successful launch of the first very high resolution (VHR) satellites capable of capturing panchromatic imagery of the land surface with ground sample distance even lower than 1 m (e.g. IKONOS in 1999 or QuickBird in 2001) marked the beginning of a wholly new age in remote sensing. On January 4, 2010, images of WorldView-2 were placed on the market. Possibly it is the most sophisticated commercial VHR satellite currently orbiting the Earth and the exploitation of its data poses a challenge to researchers worldwide. Moreover, the practice of under plastic agriculture had a great development in the Mediterranean area during the past 60 years, especially in Almeria, acting as a key economic driver in the area. The goal of this work is the automatic greenhouse mapping by using Object Based Image Analysis (OBIA). The required input data will be a pan-sharpened orthoimage and a normalized digital surface model (nDSM) for objects, both products generated from a WorldView-2 stereo pair. The attained results show that the very high resolution 8-band multispectral and the nDSM data improve the greenhouses automatic detection. In this way, overall accuracies higher than 90% can be achieved.