Radar target recognition method based on physical-statistical model

A novel radar target recognition method based on physical-statistical model is proposed, which uses sequences of full-polarization high resolution radar range-profiles (HRRP). The HRRP of target is modeled as a Middleton's Class A non-Gaussian distribution through a physical consideration. It is found that the model parameters describe the scattering physical characteristics of radar target and can be chosen as feature vector for radar target recognition. The Parzen window is adopted to estimate the probability distribution function of model parameter and the classifier is designed via Bayes theorem. The approach is applied to the classification of measured data. Numerical results have shown that the approach can extract the features which are not sensitive to target orientation and are effective to radar target recognition.