This paper proposes a method for reconstructing partially occluded faces based on a 2D morphable model. Faces are modeled by linear combinations of prototypes of shape and texture. With the shape and texture information only from an unoccluded region, we can estimate optimal coefficients for linear combinations of prototypes of shape and texture. In such an over-determined condition, where the number of points in the unoccluded region is greater than the number of prototypes, we find an optimal solution using a simple least square minimization method. Our experimental results show that reconstructed faces are very natural and plausible like real photos.
|Number of pages
|Proceedings - International Conference on Pattern Recognition
|Published - 2002
ASJC Scopus subject areas
- Computer Vision and Pattern Recognition