| Hand gestures are a key modality in the interface and a lot of information on whether the user is or is not addressing the interface might be obtained from other modalities such as gaze direction and body posture. However, I am interested to find out if this distinction can be made solely based on the input from users hands gesturing. I am developing an algorithm capable of automatically segmenting continuous gesturing into discrete gestures that are either meaningful or meaningless for the system. Meaningless gesturing can be called a null-state in human-computer interaction, analogue to letting go of the mouse. In a sense, I will teach the computer system when not to listen to its user in a purely gesture interface. |
