Gesture recognition algorithm using morphological analysis

Abstract

Recently, research into computer, vision-based methods to recognize gestures are widely conducted as means to communicate the volition of humans to a computer. The most important problem of gesture recognition is a reduction in the simplification treatment time for algorithms by means of real-time treatment. In order to resolve this problem, this research applies mathematical morphology, which is based on geometric set theory. The orientations for the primitive shape elements of hand signal shapes acquired from the application of morphological shape analysis include important information on hand signals. Utilizing such a characteristic, this research is aimed at suggesting a feature vector-based, morphological gesture recognition algorithm from a straight line which connects central dots of major primitive shape elements and minor primitive shape elements. It will also demonstrate the usefulness of the algorithm by means of experimentation. © 2013 Springer Science+Business Media Dordrecht(Outside the USA).