Gesture recognition with a DataGlove

An experiment was conducted to investigate gesture recognition with a human hand manipulating the DataGlove, an electronically instrumented glove which provides information about finger and hand position. A total of 22 gestures in three classes were investigated. The first class contained gestures which only involved finger flexure. The second class contained gestures which required both finger flexure and hand orientation. The third class of gestures required finger motion in addition to flexure and orientation. Only four sensors were necessary to positively identify specific gestures from groups of up to 15 gestures. The results show the specific number of sensors required to positively identify a gesture from a group. This depends on the number of gestures in a group, as well as the class of gestures