Robust Background Identification for Video Editing

Extracting background features for estimating the camera path is a key step in many video editing and enhancement applications. Existing approaches often fail on highly dynamic videos that are shot by moving cameras and contain severe foreground occlusion. In this talk, we present practical methods that can reliably identify background features in complex video, leading to accurate camera path estimation and background layering. Our methods significantly outperform other alternatives, and can be directly used to improve common video editing applications such as stabilization, compositing and background reconstruction. We also show a novel application benefited from the robust background identification and separation, temporal coherent video generation.