Design and Implementation of Virtual Host Based on Machine Learning Algorithm

Abstract

The research of face modeling and animation technology has a history of 30 years. With the development of virtual reality, video entertainment, videophone, human-computer interaction and other applications, this research field has attracted more and more attention. Based on machine learning algorithm, this paper studies the personalized 3D face modeling and real-time rendering of virtual hosts. The research lines of this paper are the establishment of neutral face model data structure, face feature extraction, improvement of personalized face model matching and real-time rendering in scene roaming. A realistic 3D facial lip animation system based on MPEG-4 standard and driven by voice files is designed and implemented. The system is easy to operate, versatile and efficient, and meets the needs of real-time voice-driven lip animation.