Millimeter-wave radar-based algorithm for multi-person identification

Millimeter-wave radar-based identification technology has received extensive academic attention due to its advantages of passivity and nonintrusiveness. However, current research mostly focuses on single-target vital signs identification. And research on multi-person target identification is very few. Moreover, existing algorithms are unable to accurately monitor and separate multiple human targets in a scene, resulting in a low accuracy. To address these problems, this paper proposes a multi-person identification algorithm based on vital signs. The algorithm combines the minimum variance beam forming technique, the cell average constant false alarm rate and the k-means clustering method to extract the single person vital sign signals from the multi-person target signals. Next, the algorithm designs a multiscale fusion network model to extract features from phase one-dimensional waveform signals and two-dimensional short-time Fourier transform spectrograms, and then integrates them. Finally, the purpose of identity recognition is achieved through classification. Simulation experiments show that the proposed algorithm effectively separate multi-person targets and achieve a higher recognition accuracy than that of traditional algorithms.