Improving the radar target classification results by decision fusion

The need for decision fusion appears when the classification results obtained using only one target signature are poor and have to be improved. Our idea was to use four target signatures corresponding to various polarization combinations in order to make the final decision more robust and more effective. The fusion method is based on the Sugeno's fuzzy integral and has some important advantages over the traditional methods like the vote technique. The method is validated on both synthetic and real data obtained in the anechoic chamber of ENSIETA from Brest, using four scale reduced target models. Some perspectives are finally presented for integrating also the radar imagery results into the fusion process in order to better assist the human operator to make the most appropriate decision. The main application of our research work is related to the enhancement of the airborne traffic surveillance by avoiding any confusion and increasing the flight safety.