From watermark detection to watermark decoding: a PPM approach

The possibility of increasing the payload of 1-bit watermarks through PPM encoding is analyzed from a theoretical point of view. By relying on the performance of the 1-bit watermarking system at the basis of the augmented payload algorithm, a method to upper bound the bit error probability both in the absence and presence of attacks is described. The possibility of giving an assessment on whether a given image is watermarked or not is also taken into account. Particular attention is paid to validate the theoretical analysis through experimental results. Experiments also permit to extend the analysis to cases which deviate significantly from the mathematical models. A method to speed up the recovery of position-encoded information is also described, which makes it feasible the optimum recovery of position-encoded information even for non-additive watermarking.