The application of photometric stereo to 3D facial surface measurement

Abstract

Computer graphics techniques have been extensively applied in the medical field, especially for three dimensional visualization and surgical simulation.In dental and plastic surgery, surface for the humanface is only concerned. Computed tomography(CT) and magnetic resonance imaging(MRI)may not be appropriate modalities for use in 3D surface reconstruction. Photometric stereo method is a method with the advantages of fast, cheap, and simple.It is a technique to make use of the shading information to reconstruct shape from several images taken under different directions of incident illumination.In this research , aphotometric stereo system is built and applied to measure the 3D shape of human face. From the experimental results, it is demonstrated that the human faces would be near lambertian with the help of the makeup used to remove the specularity. In addition , four images are obtained to recover the lost information in shadow areas. The near point light source model is also proposed for correction with an iterative scheme. Above all, it is shown the reconstruction accuracy of this system is better than the conventional Photometric stereo method