The effect of quality scalable image compression on robust watermarking

The effect of quantization based compression in quality scalable image coding on the robustness of wavelet based watermarking is presented. The non-blind direct modification type watermarking is considered for the analysis presented in this paper. First we present the analysis considering any quantization parameter and then restrict it to the integer powers of two to model the bit-plane discarding based quantization used in quality scalable coding, such as JPEG2000. This work assumes that the watermark embedding and compression uses the same wavelet filters. The derived model shows the relationship between the modified coefficients and the quantization factor, which is then used to obtain the conditions for the correct watermark extraction under compression. Based on this analysis, one can select wavelet coefficients for embedding the watermark in a manner that the correct watermark extraction is possible for a given quantization level. The results show higher robustness when this model is used for embedding the watermark. The paper also evaluates the performance of the model when the watermark embedding and the compression wavelet filters are not the same.