An Adaptive Compressed-DCT-Domain Watermarking

The robustness and flexibility of some algorithms based on the relationship of adjacent DCT blocks are restricted to some extent since the embedding intensity is usually not easy to be adaptive. To solve this problem, in this paper according to the principle of the maximum possible quantization error, a normalized masking deduced from Watson´s model is used to fulfil adaptive watermarking with more flexible scheme. Since based in compressed DCT domain, the proposed strategy could be directly applied in JPEG and MPEG watermarking. In addition, the watermark signal is preprocessed and is extracted without the original image. Experimental results indicate that the proposed algorithm is resistant to common image processing and noise (e.g. 3 X3 mean filter, JPEG compression)