Model-based face image coding using face prototyping and H.264
Jaroslav Polec, Sandra Ondrušová, Lenka Krulikovská · Proceedings ELMAR-2011 · 2011
Static and dynamic image coding is necessary for transfer via communication network. Digital image representation requires large numbers of bits. In this paper low bit-rate human face image coding using triangulation is presented. Human faces modeling algorithm uses an available database of average faces for face synthesis. An average face is created by image registration on 2D model of human face. Mentioned method is called a model-based coding. Strong correlation between original face and synthesis face allows minimizing of bit-rate. We have implemented the proposed approach into the images coding in collaboration with standard H.264.