Retrieval system to generate facial expressions using chaos

Abstract

An image generation method is introduced that provides creativity support for image generation. This method is driven by chaotic dynamics, employing a facial expression model constructed on fuzzy associative memories. User's creativity support regarding expressions is provided by clarifying the image through repetition of the following procedures. First, a user is shown a facial expression candidate by chaotic retrieval from a fuzzy associative memory network. The user then selects an appropriate candidate and changes the external input to the network. Computer simulation results for actual facial expression data show that this method is an effective creativity support. A leaning method is also described, in which facial expressions obtained from creativity support are memorized in the facial expression model.<>