Periodic signaling scheme in oblivious data hiding

In the oblivious watermarking, information is extracted without reference to the original host signal, whose energy is much larger than that of the watermark signal. Suppressing host noise is a great concern. Some non-linear embedding methods, e.g. Quantization Index Modulation (QIM) can greatly suppress the host noise. In this method, the signaling is periodic. In this paper we analyze the Maximum Likelihood (ML) detector for the QIM signaling scheme, and compare the error probability with hard decision detector. Based on it, a new periodic signaling scheme is proposed and its optimum and suboptimal detection is derived. Through analysis and simulation, we demonstrate its advantage in term of PE-SNR over the existing methods.