A natural hand gesture system for intelligent human-computer interaction and medical assistance

This paper presents a novel hand gesture system for intelligent human-computer interaction (HCI) and its applications in medical assistance, e. g. intelligent wheelchair control. The hand gesture vocabulary in the system consists of five key hand postures and three compound states, and its design strategy covers the minimal hand motions, distraction detection and user-friendly design. The experiment results show that the designed lexicon is intuitive, ergonomic, and easy to be remembered and performed. The system is tested in both of the indoor and outdoor environments and shows the robustness to lighting change and users' errors. The proposed intelligent HCI system can run in real-time and offers a natural and efficient interface for people with disability in their limbs to communicate with robots.