A New Approach Dedicated To Real-Time Hand Gesture Recognition.
Nguyen Dang Binh, Shuichi Enokida, Toshiaki Ejima · 2006
We introduce a new Pseudo 2-D Hidden Markov Model (P2DHMM) structure dedicated to the time series recognition (T-ComP2DHMM). The T-P2DHMM allows it to do temporal analysis, and to be used in large set of hand gestures movement recognition systems in unconstrained environments. Additionally, robust and flexible hand gesture tracking using an algorithm that combines two powerful stochastic modeling techniques: the first one is pseudo two dimension hidden Markov model (P2DHMM) and the second technique is the wellknown Kalman filter. Our work also present a feature extraction method based on the joint statistics of a subset of DCT coefficients and their position on the hand. Using feature extraction method along with the T-ComP2DHMM structure was used to develop a complete vocabulary of 36 gestures including the America Sign Language (ASL) letter spelling alphabet and digits. The results are effectiveness of the approach.