ARE CLASSIC FORENSIC TOOLS EFFECTIVE ON SATELLITE IMAGERY?

Satellite images are becoming an increasingly important part of our world. Such images are used to forecast the weather, track green house gas emissions, monitor agricultural crop health, and many other applications. Such advances are possible thanks to the free availability of a large number of satellite images. Satellite imagery now plays a key role in many areas, including external security. In this context, it is necessary to question the reliability of this data. Can the authenticity of a satellite image be guaranteed? How can one protect oneself against an entity wishing to hide illegal military material or, conversely, to incite action against another entity by falsely suggesting that it possess such material? If the forensic analysis of photographs has attracted a great deal of academic interest in recent years, this is not yet the case for satellite imagery. In this paper, we propose a methodology to create a very simple but interesting dataset to test the performance of state-of-the-art forensic methods on pristine and manipulated satellite images. Despite the strong performance of such algorithms, satellite images require special attention due to the nature of the images themselves.