Classification for Spatial Objects with Micro-motions Based on Dual CNN
Xuguang Xu, Lixun Han, Xiaojun Zou, Cunqian Feng · 2024
To enhance the precision of micro-motion objects classification under low signal-to-noise ratio(SNR), this paper introduces a micro-motion classification method for spatial objects based on dual CNN (Convolutional Neural Network). This approach takes the noisy MDS(micro-Doppler signature) image as input and output the category of its micro-motion. Firstly, to suppress the noise while preserving the MDS of weak scattering centers, one denoising CNN is designed. Second, a ShuffleNet-based CNN is developed to undertake the classification task. Finally, experiments are conducted and results demonstrate that even in low SNR conditions, our proposed method outperforms several existing methods.