Spatio-Temporal Consistency for Point Cloud Sequences Using Edge-Aware Scene Flow

We present an approach to add spatio-temporal consistency to point cloud sequences captured through a wide baseline camera setup. The system uses edge-aware scene flow to understand the movement of the point across a free-viewpoint video scene and a Hausdorff distance quality metric to reduce noise and improve density in each point cloud.Results show that the system produces temporally consistent point clouds, not only reducing errors and noise but also re-covering details that were lost in the frame-by-frame dense point cloud reconstruction. The system is successfully tested in sequences that present both static or dynamic cameras.