Maximum a-posteriori adaptive masking for clutter suppression in automatic radar target recognition

To suppress the effect of clutter in automatic target recognition we propose a maximum a-posteriori adaptive mask that isolates the target scattering centres in the data. This improves the quality of the construction of the target feature vector which results in an improved classification performance. We evaluate the technique on a closed-set identification task using targets from the MSTAR database and, on a 10-target forced-choice experiment, obtain a recognition error rate of only 1.2%. We also conduct an open set identification experiment and show that the technique can achieve an equal error rate of 1.82%.