Scholarlink

This paper presents a robust digital image watermarking method that applies Quantization Index Modulation (QIM) approach directly to the largest singular value ( \(S_0\) ) of \(4 \times 4\) SVD blocks in the \(\textrm{LL}\) subband of the Discrete Wavelet Transform (DWT) on the Y channel. The method is designed to address the lack of controlled comparisons between wavelet-domain and spatial-domain SVD–QIM watermarking using identical quantization rules. To ensure fairness, two spatial SVD–QIM baselines are constructed by applying the same QIM( \(S_0\) ) embedding rule to (i) the Y channel and (ii) the Red channel using \(8 \times 8\) SVD blocks. Experimental results on \(512 \times 512\) images show that the proposed DWT–SVD QIM scheme achieves higher imperceptibility (PSNR > 40 dB) and significantly improved robustness under JPEG compression, Gaussian noise, and blurring attacks. These findings demonstrate the effectiveness of combining wavelet-domain low-frequency localization with singular-value quantization for secure and resilient watermarking applications.

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

A Robust DWT–SVD Watermarking Method Using QIM on the Largest Singular Value: Comparative Evaluation on Y and R Spatial SVD Embedding

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Hoang Thi Thu Lan,
  • Ta Minh Thanh,
  • Hoang Phan Binh

摘要

This paper presents a robust digital image watermarking method that applies Quantization Index Modulation (QIM) approach directly to the largest singular value ( \(S_0\) ) of \(4 \times 4\) SVD blocks in the \(\textrm{LL}\) subband of the Discrete Wavelet Transform (DWT) on the Y channel. The method is designed to address the lack of controlled comparisons between wavelet-domain and spatial-domain SVD–QIM watermarking using identical quantization rules. To ensure fairness, two spatial SVD–QIM baselines are constructed by applying the same QIM( \(S_0\) ) embedding rule to (i) the Y channel and (ii) the Red channel using \(8 \times 8\) SVD blocks. Experimental results on \(512 \times 512\) images show that the proposed DWT–SVD QIM scheme achieves higher imperceptibility (PSNR > 40 dB) and significantly improved robustness under JPEG compression, Gaussian noise, and blurring attacks. These findings demonstrate the effectiveness of combining wavelet-domain low-frequency localization with singular-value quantization for secure and resilient watermarking applications.