| 摘要 |
A method for automatic reconstruction of three-dimensional human face shape according to series of photographs relates to information technologies, in particular to computer vision technologies, and can be used as component of systems for face computer identification, creation of virtual actors etc. The method is based on that the approximation of three-dimensional human face shape as linear combination of some fixed surfaces is performed by photograph series, and estimation of shot angle and face lighting parameters is performed for each photograph. This enables to automatically reconstruct three-dimensional human face shape by one photograph as opposed to known methods. |
