| Multi-Touch technology provides a successful gesture based Human Computer Interface. The contact and gesture extraction and recognition algorithms of this interface are based on full hand function and, therefor, are not accessible to many people with physical disabilities. This research extends this interface to those with disabilities. Specifically we propose two approaches to extract, estimate and recognize hand motion on the Multi-Touch Surface. Our method can extract not only finger motions, but also the motions of palm and fist, which is especially useful for the users with physical disabilities.; In this thesis, first, we design a set of command-like gestures for users with limited range and function in their digits and wrist. Then we present two different methods for hand motion feature extraction: one is to extract trajectory and angle features; the other is to extract the motion vector field by applying the Block Matching Algorithm. According to these different methods, we apply neural network and Affine model based recognition method for the recognition phase, respectively. For both cases, the analysis and numerous experiment results are illustrated. |
