A Low-complexity Hand Gesture Recognition Method Based on 2D-Trajectory Feature Using Bistatic Radars
Luntao Zhuang, Zhaocheng Yang · 2021 CIE International Conference on Radar (Radar) · 2021
Radar-based hand gesture recognition is robust to the line-of-sight and lighting environments, and has the advantages of non-contact, strong penetrability and protection of user privacy. It is becoming one of the hottest topics in the area of gesture recognition. Most of the existing radar-based hand gesture recognition methods are based on images such as time-distance map and time-frequency map, and have the problems of large model and high computational complexity. To handle these problems, this paper proposes a low-complexity hand gesture recognition method based on two-dimensional trajectory feature using bistatic radars. First, signal preprocessing and target detection are carried out on the radar data. Then, the adaptive order polynomial fitting method is used to fit the centroid and two-dimensional trajectory is then synthesized. After that, a lightweight variable-length long short term memory (LSTM) is designed for identifying hand gestures. The experimental results show that the average recognition rate of the trained people and the untrained people are 98% and 92.0% respectively.