Feature Extraction of Space Targets via Variational Mode Decomposition

Accurately and quickly identifying the attitude motion pattern of space targets is of great significance for space target surveillance. Aiming at the feature extraction for radar cross section (RCS) sequences, a new method based on variational mode decomposition (VMD) and normalized energy is proposed. First, the RCS sequences of space targets are processed by VMD to obtain intrinsic modes. Then, through calculating the energy of modes and normalizing them to the standard amplitude, feature vectors of normalized energy are obtained. The recognition effect is verified with back propagation (BP) neural network classifier. The experimental results indicate that the proposed method can effectively identify the three-axis stabilized and rotating space targets, and the accuracy rate is over 90%, which is obviously higher than the traditional statistical parametric feature extraction method.