Dynamic hand gesture synthesis and animation using image morphing technique

Abstract

Hand gesture sequence or video plays important role in various fields. In this paper, we present a method for regenerating original hand gesture sequence with the help of few frames, which are termed as key frames, which can be extracted using Hausdroff distance. The proposed technique involves extraction of quantitative description for every sign in the dictionary from a typical gesture sequence representing the sign via calculation of hand parameters, like position of fingers, location of metacarpo-phalangeal joints of fingers, and the bent angles of the fingers, for every pose in a gesture sequence. During sign regeneration, a desired gesture sequence is created by animating the corresponding key gesture frames using the extracted information. In order to bring about a natural look, we propose to generate the hand model through image morphing technique. In our experiments, we create gesture animation with hand pose that looks more or less like the real gesture thereby demonstrating the efficiency of the proposed system.