Adaptive Blind Watermarking Algorithm Against Regional Attacks of Digital Images

To enhance the algorithm's ability on resisting against regional attacks of digital images while ensuring the quality of the watermarked images, a secure, self-adaptive and robust digital blind watermarking algorithm was presented in this paper. A conception of random quaternary group was introduced, that was to select three 8×8 size blocks from each random quaternary group of the host image for hiding the redundant watermark bits and to hide the authentication watermark bits with the remaining blocks. According to the blocks' brightness-texture values, the embedding quantization steps of the redundant watermark bits and the embedding positions of the authentication watermark bits was selected adaptively. A denoising process was adopted for the extracted watermarks to improve the visual quality of the extracted watermark images. Experiments results show that, the proposed algorithm can realize the watermark blind extraction with high security and good watermark transparency, and show a high robustness on resisting conventional regional attacks and joint attacks of digital images.