Deformation invariant pattern classification for recognising hand gestures

A three stage self-organising neural network architecture has been developed to perform recognition of static hand gestures from images. Feature extraction is performed on grey-scale images by the primary stage. The second and third stage perform the recognition process. Images not recognised generate new classes by adding neural components into the second and third stage of the network. By the additional use of a hypothesis testing mechanism the network can be made to perform with no misclassifications. The network is successfully applied to a set of hand gestures by selecting network parameters according to a set of heuristic rules. The control over classification and the effects of the hypothesis testing mechanism are demonstrated using two contrasting methods of image presentation.