Real - time Two Hand Gesture Recognition with Condensation and Hidden Markov Models

This paper presents a technique for two-hand gesture recognition. There are two main processes performed to recognize hands, i.e. hand tracking and gesture cognition. In the hand tracking process, a condensation density propagation (Condensation algorithm) is used to localize and track hands (centered at the center of palms) when they are moving. Hidden Markov Model is implemented in recognition part to understand the gestures performed by human. There are 8 gestures used in this study. The experiment is conducted on the collected data (of 8 hand gestures). The results show that the proposed technique provides a promising results, achieving 96.25%.