An HMM Based Gesture Recognition for Perceptual User Interface

摘要

This paper proposes a novel hidden Markov model (HMM)-based gesture recognition method and applies it to the HCI to control a computer game. The novelty of the proposed method is two-folds. First one, the proposed method uses a continuous sequence of human motion as an input of HMM, instead of isolated data sequences or pre-segmented sequences of the data. The other one, it performs both segmentation and recognition of the human gesture automatically. To assess the validity of the proposed method, we applied the proposed system to a real game, Quake II, and then the results demonstrate that the proposed HMM can provide very useful information to enhance the discrimination between the different classes and reduce the computational cost.