A video watermarking method to objects robust against various attacks

摘要

We propose a video watermarking method in an object region, which is robust against RST (rotation, scaling and translation) and image filtering attacks. Our method embeds data in multiple video frames for robust extraction. It extracts SIFT features from a video frame and match them with those of its template to detect the object region. The proposing method enables a novel service in which product information is added to its corresponding object region and users can extract it efficiently from video using their mobile devices.