Shape from shading using linear approximation

Ping-Sing Tsai, Mubarak Shah

Abstract: In this paper, we present an extremely simple algorithm for shape from shading, which can be implemented in 25 lines of C code∗. The algorithm is very fast, taking 0.2 seconds on a Sun SparcStation-1 for a 128 × 128 image, and is purely local and highly parallelizable (parallel implementation included). In our approach, we employ a linear approximation of the reflectance function, as used by others. However, the main difference is that we first use the discrete approximations for surface normal, p and q, using finite differences, and then linearize the reflectance function in depth, Z(x, y), instead of p and q. The algorithm has been tested on several synthetic and real images of both Lambertian and specular surfaces, and good results have been obtained.