Non‐Imaging Gesture Recognition Based on Complex Representation of Vortex Waves Empowered by Machine Learning

ABSTRACT Non‐imaging target recognition by analyzing scattered waves is of vital application importance in various scenarios such as radar detection, automated systems, and life activity monitoring. Vortex wave features a helical phase structure which can be decomposed into infinite plane waves, thereby enabling information‐rich detection. Here, we develop a non‐imaging target recognition platform based on microwave vortex beams, which includes modules for target feature extraction and machine learning algorithms. A complex representation is proposed to fully characterize the amplitude and phase information of the scattered vortex waves, and a neural network (NN)‐based machine learning algorithm is used to extract the embedded information. The recognition performance is verified by experiments in distinguishing 12 different gestures from five individuals. The recognition accuracy can reach 100% for the single‐individual case and 99.1% for the cross‐individual case, completed in 0.48 and 0.117 ms, respectively. These findings offer a convenient, fast, and reliable approach for target detection and may promote broad applications in radar systems.