L1-based photometric stereo via augmented lagrange multiplier method

Recently, the sparsity model has been applied to photometric stereo by modeling non-Lambertian artifacts as sparse components. As one of these efforts, we present l₁-based photometric stereo for the non-Lambertian corruptions. A solution method was derived using the Augmented Lagrange Multiplier (ALM) method, which effectively solves the constrained problem by solving the sub-problems for surface normal and sparse corruptions iteratively. Experiments demonstrate the applicability of our method by comparing with the Least Square method and the l₁ baseline method.