Dr Gemine Vivone, a scientist at CMRE, presented the paper “Multiresolution Analysis and Component Substitution Techniques for Hypespectral Pansharpening”, co-authored with Rocco Restaino, Giorgio Licciardi, Mauro Dalla Mura and Jocelyn Chanussot, at the International Geoscience and Remote Sensing Symposium held in Quebec City in July 2014. The paper won the IEEE Geoscience and Remote Sensing Symposium Prize Paper Award 2015 for a very significant contribution to the field of endeavour in the IEEE GRS. The award-winning research was developed at the University of Salerno, just before Dr Vivone joined the CMRE.
In the last several years, data-driven science has grown to support many scientific and societal challenges. These challenges, include environmental monitoring (plant canopies, chlorophyll, etc.), climate change, and the management of natural hazards and disasters (fire detection, hurricane detection, etc.) call for novel technologies that will influence the design of relevant applications. In particular, satellite and airborne-based remote sensing play a critical role. With the recent exponential growth of operating sensors, various technical challenges emerge due to the increase availability of data sets and their heterogeneity. Thus, there is a need to integrate multiple data and knowledge related to the real world into a consistent, accurate, and useful representation, leveraging their complementary properties. The latter forms a key scientific challenge, referred to as “data fusion”.
A challenging issue in data fusion is the so-called “pansharpening”. It aims at fusing a multispectral image (i.e. tens of spectral bands with low spatial resolution) and a panchromatic image (i.e. a unique spectral band usually ranging from 0.6 μm to 1 μm with high spatial resolution), featuring the result of the processing with the spectral resolution of the former and the spatial resolution of the latter. Vivone’s paper goes beyond the literature extending several classical pansharpening techniques belonging to the multiresolution analysis and component substitution families to be used to fuse hyperspectral images (i.e. hundreds of spectral bands with low spatial resolution). The final synthetic products having high spatial and spectral resolutions can be exploited for both visual inspection and to feed further signal processing algorithms for environmental monitoring, management of natural hazards, etc. The experimental results, conducted on two real datasets acquired by the Hyperion/ALI and CHRIS-Proba/QuickBird satellite sensors, point out the better performance of multiresolution analysis techniques thanks to a higher spectral consistency of the final products, which is a desirable feature when hundreds of spectral bands have to be fused.
Dr Vivone has a broad interest in multi-sensor data fusion and data-driven science. He currently works in the Maritime Security Programme investigating multi-sensor fusion challenges for maritime surveillance. He has been heavily involved in developing algorithms to combine data using CMRE’s low-power, high resolution radar sensor network and has developed extended target tracking techniques to support vessel classification using this system. He has also worked extensively on an over-the-horizon radar capability for vessel tracking, including the use of knowledge-based techniques that incorporate information on historical vessel routes to improve tracking performance.