A Study Of Human Computer Interaction Based On Dynamic Gesture Recognition

Dynamic gesture recognition is considered as a sound solution to the Human Computer Interaction (HCI). In this regard, the hand modeling and hand movement tracking, which are two key components of the dynamic gesture recognition technology, have been thoroughly discussed.Based on subspace division, which divides the state space of the target into deformation subspace and displacement subspace, a new Particle Filter algorithm based on Mean Shift has been established. Further to this new algorithm, enhancement was made by introducing the L~2 distance as the measurement for observation. By taking the enhancement, the real time performance of the algorithm was improved.An observation modeling framework enabling two different observatory features information fusion has been established and embedded to the captioned algorithm. The new framework combines the edge contour features and color histogram features. In this connection, the robustness of the algorithm has been improved.By integrating the above observatory model, the new Particle Filter algorithm and a simple gesture interpretation algorithm, a Human Computer Interaction Control Platform has been setup and its effectiveness has been verified through varied experiments.