Robust Watermark Detection from Quantized MPEG Video Data
D. Simitopoulos, Alexia Briassouli, Michael Gerassimos Strintzis · Auerbach Publications eBooks · 2006
This chapter presents a novel approach for robust detection using the quantized domain data of Moving Picture Experts Group (MPEG) compressed video frames. The embedding can be performed either as part of the MPEG coding of live or recorded video or on precompressed MPEG bit streams by partly decoding and reencoding them. A realistic and reliable watermarking scheme needs to keep the watermark imperceptible and at the same time make it as robust as possible, so it can remain detectable even in the presence of attackers. However, in general, the Laplacian scheme exhibits better overall performance, proving to be robust for the original and also for the attacked data. Most watermarking systems aim to extract or detect a watermark without the use of the original host signal and without knowledge of the hidden watermark. Watermark embedding in the quantized transform domain is of particular importance for video watermarking.