Polarimetric Radar Target Classification Based on Decision Tree

Abstract

Due to advances in science and technology, military weaponry have been progressively improved, the current battlefield has grown more complex, and the issue of identifying aircraft targets in air warfare has gained more attention. Over time, operational needs have become more unmet by conventional radars. To successfully conduct military operations, radars that can offer more comprehensive target information are urgently required. Target classification and identification can be accomplished by using polarization radar, relying on polarization features to augment the information of the irradiation target in the polarization dimension, paired with the decision tree approach of machine learning. In this study, two scaled models of airplanes with various military functions are created and put through simulations before having their polarization characteristics described using the Poincare sphere. To perform the classification and identification of the two types of aircraft models, a classification decision tree is built and combined with the CART method. The simulation results demonstrate that the polarized feature information of the decision tree has a 94% success rate for target categorization based on the CART algorithm.