Radar target recognition based on peak location of HRR profile and HMMs classifiers

The location information of prominent points backscattering, extracted from the high resolution radar echoes by the Relax algorithm, is used as the feature for radar target recognition. The sensitivity of high range resolution (HRR) profiles to the LOS of radar yields an idea that the multiaspect radar echoes can be modeled by a hidden Markov model (HMM); it gives the spatial-temporal information of a target. Several HMMs are cascaded in a chain to model the pose of a flying objective and is used for automatic target recognition in a class-attributed and aspect-attributed manner. Computer simulations with radar data show that for a close-set recognition, an averaged recognition rate of 96.77% for aspect-recognition and 99.80% for class-recognition is achieved, respectively; that for a open-set recognition, the averaged class-recognition rate reaches 81.25% under a maneuver condition.