Aiming at the problem of how to ensure the invisibility of watermarking while improving its resistance to attacks, a digital watermarking algorithm combining Discrete Wavelet Transform (DWT), Discrete Cosine Transform (DCT) and Bacterial Foraging Optimisation (BFO) is proposed in the paper. Firstly, the original image is processed by DWT to decompose the image into multiple subbands containing different frequency information. Then, the low-frequency subbands are selected for DCT chunking operation. Finally, BFO is used to optimise the embedding position and embedding strength of the watermark, which can find the optimal solution through iterative search. It is experimentally verified that after the image is attacked by Gaussian noise, the PSNR of the watermark-containing image is between 28 and 34, indicating that the embedding of the watermark does not have much impact on the original image, and the overall NC values of the extracted watermark are above 0.9, which gives the method good robustness and invisibility.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Research on Digital Watermarking Embedding System Based on DWT-DCT-BFO

  • Minying Zhou,
  • Peng Liu,
  • Zhongrun Lv,
  • Yujuan Yao,
  • Sicheng Zhou,
  • Denghui Zhong

摘要

Aiming at the problem of how to ensure the invisibility of watermarking while improving its resistance to attacks, a digital watermarking algorithm combining Discrete Wavelet Transform (DWT), Discrete Cosine Transform (DCT) and Bacterial Foraging Optimisation (BFO) is proposed in the paper. Firstly, the original image is processed by DWT to decompose the image into multiple subbands containing different frequency information. Then, the low-frequency subbands are selected for DCT chunking operation. Finally, BFO is used to optimise the embedding position and embedding strength of the watermark, which can find the optimal solution through iterative search. It is experimentally verified that after the image is attacked by Gaussian noise, the PSNR of the watermark-containing image is between 28 and 34, indicating that the embedding of the watermark does not have much impact on the original image, and the overall NC values of the extracted watermark are above 0.9, which gives the method good robustness and invisibility.