Facial Expression Synthesis Based on Expressional Feature Space and Expression Decomposition

In this paper, we present a robust facial expression synthesis approach according to expression decomposition which based on the concept of expressional feature space. Decompose face image into facial features and expressional features, and locate facial features points and expressional landmarks automatically through ASM technology. We establish expressional vector space and facial vector space by set basic expressional vector and basic facial vector. Finally establish mapping from expressional vector space to facial vector space, and generate facial expression image. Experiment shows that the approach can synthesize arbitrary intensity new expressions based on the original expressions rapidly and effectively.