ShadowPatch: Shadow Based Segmentation for Reliable Depth Discontinuities in Photometric Stereo

Photometric stereo is a well‐established method to recover surface details and material properties. However, computing absolute depth by integrating surface normals from a signle view is inherently ill-defined. In general, self‐shadowing is considered an unavoidable complication to estimating normals. However, we observe that shadow boundaries correlate strongly with depth discontinuities and exploit self‐shadowing to create a segmentation into continuous surface patches. We deeply integrate photometric stereo with depth‐from‐stereo: Detecting continuous surface patches allows us to reduce the computational cost for correspondence search in depth‐from‐stereo. To speed‐up computation further, we merge segments into larger meta‐segments during an iterative depth optimization.