Content-based block watermarking against cumulative and temporal attack

This paper presents a block-selection-based video watermarking scheme that is designed to be resilient against two dangerous attacks: cumulative attack and temporal attack. We use content-based block selection to counteract cumulative attack by spreading the locations of marked blocks. The block selection algorithm also leads to a novel frame synchronization method that can effectively re-synchronize suspected video frames to their original positions. Our scheme has low computation overhead and robust detection performance for moderately compressed video.