Enhanced hand gesture recognition using continuous wave interferometric radar

Recently, radar micro-Doppler signatures have been extensively utilized for hand gesture recognition. As reported by existing works, recognition accuracy of different hand gestures is heavily affected by the aspect angle. In general, the accuracy deteriorates significantly with the increasing aspect angle. To solve this problem, we propose to utilize interferometric radar for hand gesture recognition in this paper, which is capable of providing two-dimensional micro-motions information, referred to as radial and transversal micro-motions. We record data of 9 different hand gestures in 4 aspect angles, where three empirical features are extracted from both Doppler and interferometric spectrograms and fed into support vector machine classifier for recognition. The experimental results demonstrate that hand gesture recognition using interferometric radar, 1) enhances recognition accuracy, 2) exhibits robustness against aspect angle, 3) recognizes horizontally symmetric gestures, by providing transversal micro-motion information and increasing spatial resolution.