Chapter 9 – Robust Watermarking

Publisher Summary This chapter discusses a robust watermarking method. In designing a robust watermark it is important to identify the specific processes that are likely to occur between embedding and detection. Examples of processes a watermark might need to survive include lossy compression, digital-analog-digital conversion, analog recording such as VHS or audio tape, printing and scanning, audio playback and re-recording, noise reduction, format conversion, and so on. However, robustness to a given process often comes at some expense in computational cost, data payload, fidelity, or even robustness to some other process. It is therefore wise to ignore those processes that are unlikely in a given application. For example, a video watermark designed for monitoring television advertisements will need to survive the various processes involved in broadcasting—digital-to-analog conversion, lossy compression, and so on—but need not survive other processes, such as rotation or halftoning. It also describes several general approaches to making watermarks robust. Some of these aim to make the watermark robust to a wide range of possible distortions. Others are general frameworks for obtaining robustness to specific distortions. Further, it analyze the effects of some valumetric distortions, i.e., distortions that change the values of individual pixels or audio samples. Common valumetric distortions that occur to photographs, music, and video can often be modeled as combinations of additive noise, amplitude changes, linear filtering, and/or quantization. Temporal and geometric distortions such as delay, translation, rotation, and scaling is also discussed.