Semantic Filtering for Video Stabilization

Abstract

Moving objects pose a challenge to every video stabilization algorithm. We present a novel, efficient filtering technique that manages to remove outlier motion vectors cause from moving objects in a per-pixel smoothing setting. We leverage semantic information to change the calculation of optical flow, forcing the outliers to reside in the edges of our semantic mask. After a 'content-preserving warping' and a smoothing step we manage to produce stable and artifact-free videos. View