Exploiting high-speed sequences for background subtraction

Observing the world in high speed can give rise to systems for e.g., tracking with super-human capabilities or extremely low latency human-computer interaction. However, the demand on computational power grows with the increased framerate, limiting the usefulness of such systems. In this paper we propose a background subtraction method aimed at high-speed video which enables higher level algorithms to focus on the interesting regions in the image. We show through experiments on high-speed video that where using conventional background subtraction methods in high-speed videos fail, the proposed method successfully detects the foreground in the image.