Abstract:

A detectable error may happen when the watermarked image suffered from joint photographic experts group (JPEG) compression. Different from various simulation tests, a mutual information function is defined as a criterion measuring the robustness of image quantization watermarking against JPEG. Based on a linear Gaussian analog model of JPEG compression, a calculation formula of the mutual information function is derived to evaluate the robustness of watermarking under quantization index modulation (QIM) scheme. In the simulation experiment, the mutual information function is calculated in terms of quality factors, and the middle frequency AC coefficients are selected as the host and are quantified according to watermark bit series in the blockbased DCT domain. The statistic bit error rate (BER) is derived under the QIM watermarking scheme and JPEG compression. Experiment results show the evaluation conclusion of the mutual information method is matched with that of the empiric BER against the quality factor. With the analog model, the mutual information method can evaluate and predict the robustness of QIM image watermarking under JPEG compression.