Patent attributes
Systems and methods for providing a virtual keyboard are shown and described. User gestures are captured by a camera and are mapped to spatial coordinates that correspond to the keys of a virtual keyboard. The user defines the coordinate system based on his or her range of motion and also defines the spatial dimensions of the virtual keyboard. The spatial dimensions are then scaled to provide a display image of the virtual keyboard on a TV display. Facial recognition techniques and corresponding data regarding the viewer's anatomy and previously captured reference gestures are used to interpret the viewer's gestures and determine which keystrokes are intended. A character prediction technique using the trajectory of the cursor (i.e., trajectory of entered keystrokes) is combined with language/semantic-based character prediction models to identify a next predicted character that is indicated as the user's selected character, thereby disambiguating the key selection indicated the positioning of his or her fingers relative to the virtual keyboard keys.
