ViHand: Gesture Recognition with Ambient Light

Hand gesture recognition has become increasingly important in human-computer interaction (HCI) and can support a broad range of emerging applications, such as smart home, virtual reality, and mobile gaming. During the last few years, more and more researchers are exploring ubiquitous modalities, such as radio frequency signals and acoustic signals, to enable gesture recognition. Compared with existing methods, the light-based approach leverages ambient light (daylight, lighting, etc.) to detect and recognize human gestures, which is totally non-intrusive and very convenient for daily use. In this paper, we develop a prototype system, named ViHand, to facilitate automatic detection and recognition of gestures by using ambient light. The key idea of light-based gesture recognition is quite straight forward: when moving with different gestures, the hand will shade the sensor from the light with different orders, which will generate a unique shadow pattern. ViHand uses photodiode sensor arrays to capture this unique shadow by proposing a two-step recognition approach. Specifically, we use the order in which sensors are blocked to recognize the sliding gestures. Furthermore, for the recognition of complex gestures such as digital gestures, we first establish a priori template library based on the signal characteristics caused by the motion of different gestures. Then, an improved dynamic time warping(DTW) algorithm is used to match the template, and the kNN algorithm is used for classification. We conduct a set of experiments to verify the effectiveness of our system, and the experimental results indicate that the classification accuracy of sliding gestures reaches 100%, and the accuracy of digital gestures reaches 82.3%.