Photometric Stereo With Near-field Quasi Point Light Sources

This paper presents a new computational framework for the Photometric Stereo system with near point light sources. Firstly, a quasi-point light model named QPLM is presented. with such a light model, radiance property of light source can be more precisely represented. Different with conventional point light model, not only the angular and distance attenuation factors but also the attributes of light source are considered in the proposed QPLM. Secondly, to estimate system parameters, a two-step calibration procedure is introduced. By defining the camera as world coordinate system, a multiple-sphere-based approach is used to estimate the light source positions firstly. And then, a reference-plane-based approach is applied to estimate the principle optical axis of each light source as well as its principle radiant energy. For each surface point on the target, the direction and intensity of its incident light ray can be precisely determined by the calibration parameters and the quasi-point light model. At last, 3D reconstruction of the surface with a quasi-point light source is also implemented in two steps. The target is firstly initialized as a known geometric surface, such as planar, and reconstructed via conventional surface integration means. Based on the initial 3D reconstruction result, an iterative procedure is then introduced to optimize the lighting condition for each surface point with respect to its surface depth. In the iterative process, by estimating the mean value of depth, the surface depth is projected into the physical coordinates. In the experiments, both planar and free-form surfaces including real human faces are used. And the results show that, higher accuracy surface normal as well as better 3D reconstruction quality can be obtained by the proposed approach in comparison with conventional means.