Shape from shading using three images

Methods to digitize objects are necessary in many application areas to feed the surface of existing workpieces into the computer. If the data of a surface is stored, then it can be modified by suitable programs. Moreover, machines can be controlled by it to produce this workpiece. One possibility for reconstruction of a surface is to examine the distribution of image brightness of the surface. There exist several methods using only a single image. Then, since this problem is inherently underdetermined, there must be introduced additional assumptions about the surface. In contrast to this, the algorithm presented in this paper uses three images and therefore requires fewer assumptions about the surface. We will show that the reconstructed surface converges to the surface of the object, as the resolution of the image and the number of grey colours tend to infinity.