Calibration of position and orientation for point light source synchronously with single image in photometric stereo.

Abstract

In this paper, a calibration method for non-isotropic point light source is developed, which is capable of calibrating position and orientation of the point light source with a single image synchronously. Based on the bidirectional reflectance distribution function (BRDF), the relationship between the shape of observed surface and the recorded grayscale image is firstly investigated in a more accurate way. Based on this model, a cost function is proposed to evaluate the estimation error, and a new cumulative, error-free iteration method is designed to recursively estimate the point light source parameters. The average simulation rebuilding root mean square error (RMSE) is below 0.9. The experiments are performed by calibrating a point light source in a photometric stereo measurement system, and the corresponding RMSE is below 1.2 in the unit of grayscale. Further analysis shows one image is enough to achieve the precision calibration using our proposed method.