Constructing a 3D facial model from a single face image
Lei Wen · 2015
In this thesis, we present a novel framework to reconstruct 3D face from one image with a 3D Morphable Model. 3D Morphable Models have been widely used as a tool to model 3D objects and generate images of a class of objects and to analyze them. There are many proposed methods related to it in the literatures, here we extend Blanzs 3D face synthesis method with facial feature extraction algorithm, which improve the efficiency, robustness and accuracy. We firstly use Viola-Jones classier to detect the face region from an input 2D image, and fit the face region to Active Shape Model to determine the positions of the facial feature points. Then we derive a morphable face model by transforming the shape and texture of the examples into a vector space representation. New faces and expressions can be modeled by forming linear combinations of the prototypes. Finally, we propose a reconstruction algorithm using 2D image-based registration based on facial feature alignment and PCA face models. Since our pca model contains only shape components, we adapt texture mapping techniques to obtain the final result. Although improvement is needed, our work can potentially unveil applications such as 3D face recognition, reshaping face image, and facial animation.