Shape from shading-based 3D face shape recovery fiom single 2D image

3D depth information of a face, which does not vary under different exterior conditions, provides an excellent clue to current face recognition (FR) research. Among techniques of 3D shape recovery, shape from shading (SFS) method which can recover from single 2D image is most popular. However, most existing SFS algorithms assume that objects in the scene are Lambertian reflectors and the reflectivity of the object's surface is uniform, which makes it difficult to reconstruct the face shape successfully and accurately, mainly because the face not only has a complex shape but also is composed of parts with different reflecting properties such as cheek and lip. In this paper, an improved algorithm is proposed. The strategy consists in: 1) according to face's shape and albedo symmetry, an albedo free reflectance function is derived; 2) through an eye detector, the 2D face image being measured is aligned to a 3D face model; 3) the iteration equation for computation of the depth value is obtained and the 3D depth information of the model is set as initial depth value of the face shape. Through continuously iteration until convergence, the 3D depth information of the face is achieved lastly; 4) based on the depth values calculated, the pixel-wise albedo value is also obtained. Experiments are carried out on symmetry objects and real face images. Results show that the proposed SFS algorithm can efficiently accomplish face's 3D shape recovery in less time and at less iteration.