Spread spectrum watermarking for images and video.

Abstract

This thesis describes a digital watermarking method using the principle of spread spectrum communication. This method could be used for owner authentication or for fingerprinting. Though this work concentrates on image and video watermarking, the same method could be used for audio and text (represented as an image) as well. The major challenge in watermarking is that the watermarks should withstand all types of intentional and unintentional attacks, at the same time being imperceptible. Previously proposed digital watermarking techniques were not very robust, and the watermark was easy to remove. Using Direct Sequence Spread Spectrum, it is possible to embed invisible watermarks, which are highly robust to attacks and unintentional transformations. The robustness of this scheme to different signal manipulations and transformations is studied. This thesis also covers the different aspects of watermarking and includes an information theoretic analysis of the spread spectrum watermarking for fingerprinting. Even though spread spectrum signals are very resistant to non-linear amplitude distortion and additive noise, they are not resistant to timing errors. Using a sliding correlator for instance, it is possible to detect and compensate for the loss in synchronization. Our simulation results demonstrate the robustness of the spread spectrum watermark to lossy compression, filtering, cropping, rotation, scaling, printing and scanning, multiple watermarking attacks, collusion attacks, etc.