New method for radar HRRP recognition and rejection based on weighted majority voting combination of multiple classifiers

To deal with the problems of recognition and outlier rejection in radar automatic target recognition (RATR), a novel method based on weighted majority voting combination of multiple classifiers is proposed. In both the rejection and recognition phases, the results of multiple classifiers which employ different features are combined. In the rejection part, the nearest neighbor (NN) classifier with the outputs of a classifier as the input features is utilized to identify outliers. One cost function is proposed to measure the recognition and rejection performance of each classifier. Experimental results based on the measured HRRP dataset validate the effectiveness of the proposed method.