The research on digital watermarking algorithm based on neural networks and singular value decomposition
Xuezhang Zhao, Jianxi Peng, Yunjiang Xi · 2011
The digital watermarking algorithm based on discrete Hopfield neural networks and singular value decomposition have been put forward in this paper. Currently, there are some shortcomings such as weak resistance on the attacks of geometric distortion and noise in some digital watermarking algorithm. Firstly, block-based singular value decomposition was applied for the original image to build the neural network model between the largest and other singular values which are coefficient. Secondly, the watermark information has been embedded in selecting larger contrasting image blocks. Finally, the neural networks were used for memorizing the original image and watermark information. While checking, the watermark was extracted by use of embedding watermarking image and associating original image and then the blind detection was realized. The results show that the watermark can be well extracted through the common image processing and compressing operations and the watermarking algorithm is with the good robustness in the geometric distortion.