AN EFFICIENT DCT-BASED VIDEO WATERMARKING FOR COPYRIGHT CONTROL

Abstract

Since the technology is being improved, digital media are facing challenges like copyright infringements. Digital data can be easily created, copied, processed, and distributed freely among unauthorized users. The author or owner of the data does not know that the duplicate files of his work are available on the Internet that can be accessed by anybody. The copyright laws are also not sufficient to deal with the digital data. With the aim of copyright control in transmission of digital video, this thesis presents a Discrete Cosine Transform (DCT)-based video watermarking method. Its essence is to embed the copyright related information on digital videos in such a way that it can later be extracted in case copyright violation is detected. Watermark used in copyright protection applications is information that can uniquely identify the owner, such as logo, signature, etc. For copyright protection, the watermark must be carefully embedded in the host video. A good watermarking system should not degrade the visual quality of the host video. It means that watermarks used in the copyright protection applications should not be perceptible by human eyes. It should also resist common signal processing attacks. Only when the copyright violation is detected, it should be able to successfully extract the watermark from the manipulated video to prove the ownership. In this system, watermark is embedded by only changing the luminance of the video frames. As the human visual system cannot easily detect light intensity changes in the images, the method presented in this thesis well preserves the visual quality of the watermarked videos by keeping the Peak Signal to Noise Ratio (PSNR) of more than 50dB. Moreover, this system solves the frame drop problem by repetitively embedding the watermark in all video frames. Therefore, even if some of the video frames were lost during transmission, other remaining frames can successfully be used for watermark detection. The system also provides remarkable robustness against compression attack as it is based on the Discrete Cosine Transform (DCT), which is a proven method used in Joint Photographic Experts Group (JPEG) compression. In addition to compression attacks, the robustness of the system is also tested against various kinds of signal processing attacks: such as compression, rotation, cropping, quantization, filtering, and noise addition. According to the experimental results, the system shows a great ability to preserve the watermark against those attacks.