A secure HEVC video watermarking scheme for authentication and copyright purposes

• Main Author: Elrowayati, Ali Abdulhafid
• Format: Thesis
• Language: English; English; English
• Published: 2020
• Subjects: QA76 Computer software
• Online Access: http://eprints.uthm.edu.my/883/1/24p%20ALI%20ABDULHAFID%20ELROWAYATI.pdf; http://eprints.uthm.edu.my/883/2/ALI%20ABDULHAFID%20ELROWAYATI%20COPYRIGHT%20DECLARATION.pdf; http://eprints.uthm.edu.my/883/3/ALI%20ABDULHAFID%20ELROWAYATI%20WATERMARK.pdf; http://eprints.uthm.edu.my/883/

High-Efficiency Video Coding (HEVC) becomes one of the widely deployed standards for multimedia applications. However, HEVC streams can be easily tampered by any third party, which negatively affects the authentication and copyright protection. Existing watermarking schemes used for copyright purpose are not able to protect the copyright information, especially if the hosting video encountered some intentional and/or unintentional attacks, such as recompression attack, lossy channel attacks, signal processing attacks, frame deletion attack, and image processing attacks. In addition, existing watermarking schemes used for authentication purpose are mostly suffering from the inability to detect recompression attack, especially if it uses the same quantisation parameters as the original compression. Further, existing watermarking schemes are suffering from the inability to locate tampering in videos. Moreover, some of those schemes could allow unauthorized access over an insecure channel, which is considered a serious security issue. In order to solve these issues, two HEVC video watermarking schemes are proposed; (1) a zero-fragile watermarking scheme based on sensitive watermarking zone and (2) a robust watermarking based on invariant watermarking zone. Additionally, the error correction code and cryptography techniques are applied to the watermark information to increase robustness and security over insecure channels. The first proposed scheme shows enough sensitivity to successfully detect video tampering, distinguish between intentional and unintentional attacks, and differentiate between first and second video compression at different bitrate, with accuracy improvement up to 42% compared to the-state-of-the-art schemes. Moreover, the second proposed scheme shows significant improvement; up to 8.23% of robustness against recompression attack, 95% against channel noise attacks, and 5.37% against frame deletion attack compared to the state�of-the-art schemes. Additionally, both proposed schemes are capable to maintain high visual quality, minimum bitrate increase, and high embedding capacity. Furthermore, both proposed schemes can localise tampering and prevent unauthorized access to watermarked information even over insecure channels.