A Shape-From-Shading Algorithm Using Photometric Stereo

We present a simple shape-from-shading algorithm that uses two shading images. This enables us to estimate more accurate shapes of objects without any special constraints or conditions than the existing photometric stereo algorithms. The Jacobi’s iterative method is applied to the difference between the image and the reflectance function represented as a function of three depth parameters for each of the two images, and the resulting two iterative relations are combined to get a single iterative relation. The algorithm is analytically shown to be numerically stable. Two supplementary methods are also presented to enhance its applicability. Computer experiments clearly show its usefulness and robustness to noise. Typically it takes seconds using a 1.2GHz Pentium III PC to get shapes from two 50x50 shading images.