Structure from shadow motion

Abstract

In outdoor images, cast shadows define 3D constraints between the sun, the points casting a shadow, and the surfaces onto which shadows are cast. This cast shadow structure provides a powerful cue for 3D reconstruction, but requires that shadows be tracked over time, and this is difficult as shadows have minimal texture. Thus, we develop a shadow tracking system that enforces geometric consistency for each track and then combines thousands of tracking results to create a 3D model of scene geometry. We demonstrate reconstruction results on a variety of outdoor scenes, including some that show the 3D structure of occluders never directly observed by the camera.

Time‐Lapse Photometric Stereo and Applications

“…While it is a challenging problem, the ability to handle soft shadows would be an interesting extension of our technique. In general, shadows contain useful information about scene geometry [ASP14], and leveraging this information could make our technique more robust.…”

“…Abrams et al [AHP12] presented an alternate technique that uses months of data to reconstruct Lambertian scenes. In subsequent work, they proposed using shadows to recover sparse depth from time-lapse sequences [AMP13,ASP14]. Yu et al [YYT * 13] use a small number of images captured in a single day along with light probes capturing the incident illumination to reconstruct surface normals.…”