Shape Reproduction of Object with Many-Colored Surface by Photometric Stereo Method

Abstract

Three-dimensional shapes created from photographic imagery using a photometric stereo method is restored in the present study. We used one object and one digital camera. The source of light was moved and the camera took a picture of two or more object shape data. The radiance value of the obtained photographic imagery was requested respectively. It was restored to three-dimensional geometry by calculating the normal vector from the radiance value, and combining the data of the photographic imagery. As for this method, it is well-known that application is difficult for objects that have lustre. Therefore, an improvement was attempted using spline interpolation. It has been understood that the method which reproduces it by using spline interpolation as a result is effective. Furthermore, because an error margin is caused when an object with colour variation is reproduced, an improvement was tried. First of all, the colour component of the image was extracted. Then, the image with a small change in brightness was chosen and reproduced. Consequently, a shape near the theoretical value was able to be reproduced for the reproduction of a two-colour object including white.