A Micro-Doppler Curve Parameter Estimation Method Based on the Short-Time Ambiguity Function Entropy and the Sparse Bayesian Learning

Micro Doppler is an important feature of the micro motion. Micro Doppler analysis plays an irreplaceable role in the recognition and detection of moving targets such as human motion, ground vehicles, helicopter rotors, small boats, sonar. In the micro Doppler analysis, period estimation is one of the important applications. Although the motion states (parameters) of micro motion parts are different, the period is the same. The period estimation algorithm based on cross-correlation can only estimate the period of micro Doppler curves without additional translational motion, and is susceptible to noise interference. In this paper, a short-time similarity framework is introduced for the micro Doppler analysis and parameter extraction of human body parts with additional translational motion. This is an evaluation system based on cross-correlation, which can provide more accurate period estimation for micro Doppler signal with additional translational motion. In this paper, the short time ambiguity function entropy and the sparse Bayesian algorithm are combined to estimate the period of Doppler curves of human body micro moving parts (arms, feet, legs) and the remaining parameters of Doppler curves. By using the Boulic model of the human body, the simulation experiments have verified the effectiveness of the proposed algorithm framework. Compared with other micro Doppler analysis algorithms, this algorithm framework can accurately separate micro Doppler curves of different parts in a shorter time.