Dynamic hand gesture recognition using the microwave or millimeter-wave radar sensors has become a typical technology for many human-computer interaction (HCI) applications. In this letter, a novel method is proposed for dynamic hand gesture recognition based on micro-Doppler radar signatures. The short-time Fourier transform is carried out on the raw data to obtain the time-frequency spectrogram. The time-frequency spectrograms associated with the same dynamic hand gesture are modeled by a hidden Gauss-Markov model (HGMM), and the testing gesture is recognized by the maximum likelihood criterion. Experimental results with real radar data demonstrate that the proposed method has a strong generalization ability for radar gesture recognition in the cases of low signal-to-noise ratio (SNR) and unknown users.

To improve the performance of the recently developed parameter‐dependent sparse recovery (SR) space–time adaptive processing (STAP) algorithms in real‐world applications, the authors propose a novel clutter suppression algorithm with multiple measurement vectors (MMVs) using sparse Bayesian learning (SBL) strategy. First, the necessary and sufficient condition for uniqueness of sparse solutions to the SR STAP with MMV is derived. Then the SBL STAP algorithm in MMV case is introduced, and the process for hyperparameters estimation via expectation maximisation is given. Finally, a computational complexity comparison with the existing algorithms and an analysis of the proposed algorithm are conducted. Results with both simulated and the Mountain‐Top data demonstrate the fast convergence and good performance of the proposed algorithm.

Subspace space-time adaptive processing (STAP) algorithms are able to eliminate clutter completely. However, when the number of training samples is smaller than the clutter rank, the performances of subspace STAP algorithms degrade severely due to the inaccurate estimate of clutter subspace. To remedy this problem, a novel subspace STAP algorithm is proposed. In the proposed algorithm, the entire clutter subspace is constructed by two portions. The direct portion is estimated by a conventional method from limited training samples, while the supplemented portion is constructed by some space-time steering vectors selected from an overcomplete space-time steering dictionary. Clutter suppression is achieved by projecting the data into the subspace orthogonal to the clutter subspace. Numerical results with both simulated and mountain-top data demonstrate that the proposed algorithm has superior performance in a finite-training-sample situation.