Static hand gesture recognition based on 2-D SAR imaging

Hand gesture recognition has long been a study topic in the field of Human Computer Interaction. Compared with traditional camera-based recognition systems, radar can realize dynamic hand gesture recognition at long distances and in low light conditions by exploiting the micro-Doppler (m-D) effect. However, for some static and complex hand gestures, the m-D-based recognition methods are rendered impotent. In this paper, we present a method based on 2-D synthetic aperture radar (SAR) imaging to distinguish nine kinds of static hand gestures representing the numbers 1-9. A cost-effective 77 GHz mm-wave radar is used to achieve imaging and data acquisition of different gestures. Finally, two classifiers including classic machine learning and deep learning are used to evaluate the effectiveness of the proposed method. Experimental results demonstrate that the proposed method can guarantee a promising recognition accuracy.