Finger Gesture Recognition using Wi-Fi
Oral Defence Date:
Wednesday, May 16, 2018 - 14:00
Asst. Prof. Hao Yue and Assoc. Prof. Hui Yang
Computers have become so common in our daily life, and there has been an increase in demand for Human Computer Interaction. Gesture recognition is an emerging technology for Human Computer Interaction and has a wide variety of applications like gaming and virtual reality. Previous work and methods on finger gesture recognition either use sensors that are worn by users or cameras that require proper light to identify gestures. In this project, we develop a new finger gesture recognition system with a commercially available Wi-Fi device, which utilizes the Channel State Information (CSI) of the received Wi-Fi signals at the Wi-Fi device and extracts unique patterns of the CSI to identify different finger gestures. Specifically, we first develop an environmental noise removal mechanism to mitigate any noise caused due to environmental changes. Then, we use two different approaches, i.e., Principal Component Analysis (PCA) and Eigenfaces, to create profiles for known gestures. Given a testing gesture, our system calculates the similarities between the extracted CSI pattern and the gesture profiles, and recognizes it as the gesture with the profile that has highest similarity. Our experimental evaluation shows that the system we propose can achieve about 75% accuracy with PCA and 85% with Eigenfaces. Experimental results also show that the system is robust to environmental noise as well as individual diversity.
Wi-Fi, Channel State Information, Principal component, Eigen vector