Automatic visual recognition of gestures made by motor-impaired computer users

Abstract

The use of gesture recognition as a means of Human Computer Interaction for physically disabled users is discussed. The ability of motor-impaired computer users to make distinct, recognisable gestures is not exploited by current assistive technologies. A real-time computer vision system for recognition of one- and two-handed gestures defined by such users is described. An investigation into the feasibility of real-time, unencumbered recognition of gestures defined by motor-impaired users by means of Hidden Markov Models, trained with relatively few examples is performed and reported. Different feature vectors are compared and the trade-off between accuracy and training set size is explored: an important issue for such interactively trained systems.