FULLY AUTOMATED SENTINEL-2 DATA REGISTRATION TO VARIOUS MULTISENSOR EARTH OBSERVATION IMAGING DATASETS

摘要

In this paper, we propose a deformable registrationframework for the fully automated, accurate registrationof Sentinel-2 datasets. The proposed approach is basedon a robust non-rigid registration framework under aMarkov Random Fields formulation. Efficient linear programminghas been employed for reaching the lowest potentialof the cost function. The framework has been validatedfor the registration of Sentinel-2 and various multisensoroptical and radar data like Sentinel-1, RapidEye,Landsat, Proba, Modis, etc. The performed quantitativeand qualitative evaluation demonstrated the high potentialsof the developed approach both for optical and radardata.