A sparse representation method for determining the optimal illumination directions in Photometric Stereo

•The optimal illumination configuration for Photometric Stereo.•A sparse representation of the illumination using l1 optimisation was considered.•A hemisphere structure for the sampled lights was considered.•The Lambertian model was considered.•Experiments were performed with real and synthetic data to verify the outcomes.

The analysis of surface and texture details with the help of changes in illumination direction is a key task in 3D shape reconstruction either based on Photometric Stereo, Shape from Shading or Structured Light. This paper presents a novel approach for estimating the optimal illumination directions for the accurate calculation of the surface normals, while minimising the presence of shadows and the reconstructed albedo error. The method regards a sparse representation of the illumination arrangement and estimates the light directions using l1 optimisation. The Lambertian model is considered and the theoretical development is demonstrated with experimental results.