This dissertation proposes a framework of techniques that can be used for virtual face mirroring. Virtual face mirroring is driven to perform the same actions as those of the human face by adapting a blendshape template to an image sequence of recorded expressions using an optimization integrates facial feature tracking.

The mirroring process is divided into three major steps: face reconstruction, blendshapes construction and driving facial animation. The 3d face reconstruction consists of building linear 3d morphable model (Blanz 99) and solving non-linear equations based on facial features extracted from a single image . Then we adapt deformation transfer (Sumner 04) to automatically register the built model to construct blendshape models. Finally we convert the driving facia...[ Read more ]

This dissertation proposes a framework of techniques that can be used for virtual face mirroring. Virtual face mirroring is driven to perform the same actions as those of the human face by adapting a blendshape template to an image sequence of recorded expressions using an optimization integrates facial feature tracking.

The mirroring process is divided into three major steps: face reconstruction, blendshapes construction and driving facial animation. The 3d face reconstruction consists of building linear 3d morphable model (Blanz 99) and solving non-linear equations based on facial features extracted from a single image . Then we adapt deformation transfer (Sumner 04) to automatically register the built model to construct blendshape models. Finally we convert the driving facial animation problem to minimising a distance measurement between face image and constructed blendshape models, which is a non-linear least squares problem and solved by the Levenberg-Marquardt algorithm in our work.

We build a software program to facilitate the mirroring process. The program adequately meets the goal of real-time mirroring, but the accuracy is only acceptable. In this dissertation we will discuss in detail the technical background of our problem and the results of our software program. This work can potentially suggest new possibilities of social network and human-computer interaction.