Non-Sinusoidal micro-Doppler Estimation Based on Dual-Branch Network

The fine state of targets can be represented by the extracted micro-Doppler (m-D) components from radar echo. However, current methods do not consider specialty m-D components, and their performance with non-sinusoidal is poor. In this paper, a neural network applied to signal extraction for first time. Inspired semantic line detection in computer vision, transformed into network-based time–frequency curves problem. Specifically, novel dual-branch method proposed. According property intersected multiple consisting continuous branch, crossing point branch designed obtain cross points at same addition, shuffle attention-fast Fourier convolution (SA-FFC) module proposed fuse local global contexts focus on key features. To solve error correlation problem multi-component signals, first-order parametric condition cubic spline interpolation are employed complete smooth curves. Simulation measurement results show that good robustness candidate separating intersections.