Digital Watermarking Technique For Image Authentication And Copyright Protection

Digital images and videos are more and more widely used in nowadays society. They are much more convenient in editing, storing, transmitting and usage. But unfortunately, due to the malicious modification, digital media become unsafe. Therefore, effective copyright protection and multimedia security are becoming a highlight.In this thesis, a novel approach to content authentication for JPEG images is proposed. The image authentication method is based on a semi-fragile feature. Our proposed scheme has the capability of differentiating "content preserving" modifications from "malicious" attacks and so can ascertain and locate the "malicious".Also, we proposed a robust watermarking algorithm for copyright protection. We employ channel estimation principles to improve the performance of robust watermarking. A binary symmetric channel (BSC) model of the watermark channel is introduced to more accurately characterize signal tampering. By relying on statistical Bayes decision theory, we make decisions on the extracted watermark to further improve the performance.