For the third year in a row, a Tor Vergata Earth Observation Lab and GeoInformation PhD student wins the IEEE Remote Sensing Data Fusion Contest

This year the winner team was composed by Fabio Pacifici, a Tor Vergata GeoInformation PhD student and Nathan Longbotham from the University of Colorado. The goal of the contest was to map the flooding in the Gloucester area (UK) using data from different sensors and different dates. Particularly, two ERS and two SPOT images have been made available (before and after the flood) for this change detection exercise.

The winner algorithm was (again) a NN-based algorithm (MLP) followed by a morphological erosion/dilation to fill the flooded area.

Note that another Tor Vergata Earth Observation Lab and GeoInformation PhD student, Giorgio Licciardi, already winner of the 2008 contest, ranked second in the fusion of best results. Also Giorgio's algorithm was based on Neural Networks, while also this year most of the teams used SVMs or other complex algorithms