Automatic Real-Time Generation Of Talking Cartoon Faces From Image Sequences In Complicated Backgrounds And Applications

Abstract

A system for automatic real-time generation of talking cartoon faces is proposed, which includes five processes, namely, preprocessing, facial feature tracking, image feature point transformation, speech recording, and animation generation. The processes are designed to allow the system to create cartoon faces fast enough for real-time display of the result. A method for creation of oblique 2D cartoon faces is also implemented. Two interesting applications on networks are also implemented, namely, multi-role avatar broadcasting through network, and web TV by the ActiveX technique. Experimental results show the feasibility of the proposed methods.