Blind MPEG-2 video watermarking in DCT domain robust against scaling

Abstract

Watermarking techniques that need no original information during watermark detection, known as blind watermarking, are more desirable than informed ones for practical usage and convenience in watermark extraction. A blind MPEG-2 watermarking technique operating in the DCT (discrete cosine transform) domain is presented, which is generally robust against arbitrary ratio scaling, provided that turbo codes are used for error correction. The method can be directly applied to other block-DCT-based video compression techniques. The main advantage of the scheme is its simplicity, blindness and the ability to tailor the watermark payload in a trade-off with other watermark requirements. As synchronisation is automatically obtained from the frames in the scaled video itself, a connection between the source and the target video is not necessary for watermarking detection.