Digital video watermarking of H.264/AVC
Jing Zhang · 2008
A digital video watermark is information stored invisibly in the video content to which it belongs.This is done by slightly changing the video content according to a secret pattern.While this is imperceptible during normal viewing, a decoder can detect and retrieve the watermark by using the same secret pattern.Once applied, the watermark cannot be deleted or changed.It is inextricably bound to the video content, surviving signal processing steps or data compression, but also geometric distortions.Watermarking techniques enable a variety of content protection applications.For example, the developed robust video watermarking scheme is designed to work with the upcoming digital-cinema format, and can help content owners and distributors by tracing the origin of illegal copies of movies made in the cinema, which is an increasingly occurring practice.Digital watermarking has first been extensively studied for still images.In this PhD thesis, a novel image-in-image watermarking algorithm is first studied for copyright protection.This algorithm is aimed to test the watermarking performance of the integer DCT transform used in the H.264/AVC standard, which is an approximation to the traditional DCT.Edge detection is used to extract the significant image feature for adaptive embedding and some effective post-processing approaches are utilized to increase the watermark robustness against common signal processing.The novel image-in-image watermarking scheme shows good performance and is benchmarked with StirMark 4.0.