Recurrence Plot Application of RCS Sequence for Radar Target Classification with Micro-motion

Micro-motion is of great significance for classification of spatial targets. Based on the RCS (Radar Cross Section) sequence induced by micro-motion, this paper proposes a spatial target classification method based on RP (Recurrence Plot) and CNN (Convolutional Neural Network). First, several micro-motion types of spatial targets are analyzed and thus different models are established. Then, according to the characteristics of delay time and embedding dimension, phase space reconstruction method is used to generate multi-channel and non-threshold RPs. Finally, RPs are inputted into Xception network to classify targets with different micro-motions. Compared with the original RCS sequence, RP-based method has high classification accuracy.