Research on Multi-Dimensional Feature Fusion Recognition Method of Micro-Motion Gesture based on Millimeter Wave Radar

Current research on millimeter-wave radar gesture recognition mostly focuses on gestures with larger amplitudes, which are difficult to meet the needs of human-computer interaction on the mobile terminal. There are also problems such as the inability to accurately classify gestures with a single feature, and the large amount of model calculations that makes it difficult to apply in practice. In response to the above problems, this paper proposes a micro-motion gesture multi-dimensional feature fusion recognition method based on millimeter wave radar. This paper focuses on smaller finger movements, collects the original signals of 7 micro-motion gesture movements of multiple people, obtains clustered RDI (C-RDI) and RAI (C-RAI) through micro-motion gesture data processing methods. In addition to using a single feature, range, Doppler, and angle features are combined to characterize micro-motion gestures, and a Spiking Neural Network (SNN) with fewer parameters is used to classify micro-motion gestures. Experiments show that the multi-dimensional feature fusion recognition method can achieve a recognition accuracy of 97.6%, which is 11.5% and 3.3% higher than using a single C-RDI or C-RAI feature respectively.