In: IEEE Trans. on Systems, Man, and Cybernetics, Part A: Systems and Humans, Vol. 29, No. 5, pages 514-521. 1999.
Abstract: The main characteristics of human hand gestures can be summarized by their dynamic, multi-attribute property. To utilize hand gestures as a way of interaction, it is necessary to analyze the motion patterns for each of the gesture attributes and finally to extract the whole interpretation by integrating the relevant factors across time. Previous research have shown the possibility for recognition of local aspects of hand gestures. But the global framework for finding the whole interpretation from the local aspects has yet to be provided. This paper proposes a colored Petri net model for high-level description of hand gestures. This model intercommunicates with simultaneous low-level recognizers and thus finds a whole-interpretation for the gesture.
Keywords: colored Petri nets, hand gestures, hidden Markov models, pattern recognition.