Shading Based 3D Shape Recovery in the Presence of Shadows

Shadows usually cause various problems in three-dimensional shape recovery and measurement methods. In particular shading based approaches such as shape-from-shading or the photometric stereo method produce no or wrong results if the shadows are not treated appropriately. We show how information extracted from shadows can be employed to reduce the problems caused by them. This is done for multiple light-source photometric stereo. Unlike other published work, we formulate sufficient conditions to recover locally unique surface normals from two image irradiances (intensities) and a zero-irradiance caused by a shadow. We also distinguish between self-shadows and cast-shadows. Moreover we show how much information is obtainable by using the shadow analysis.