研究目的
To introduce a secure and robust image watermarking procedure based on SVD and SFLCT that resolves the false positive problem and resists ambiguity attacks without extra steps.
研究成果
The proposed SFLCT-SVD watermarking scheme effectively combines the advantages of SFLCT and SVD to achieve high imperceptibility, robustness, capacity, and security. It resolves the false positive problem inherent in SVD-based methods without additional steps, making it suitable for applications like proof of ownership and copyright protection. Future work includes extending the scheme to color images and further optimizations.
研究不足
The SFLCT is not shift-invariant, which may affect watermark extraction accuracy. The computational complexity is O(N^3) due to SVD, which could be high for large images. The scheme is tested on grayscale images; applicability to color images is not addressed and is noted for future work.
1：Experimental Design and Method Selection:
The methodology involves combining SVD and SFLCT transforms for watermark embedding and detection. SFLCT is used for its directional multiresolution properties, and SVD for its robustness to perturbations. The embedding process applies SFLCT to host and watermark images with specific downsampling rates, modifies approximation subbands and singular values of details matrices, and uses inverse transforms. The detection process reverses these steps to extract the watermark.
2：Sample Selection and Data Sources:
Test images include Lena (512x512) and Cameraman (512x512) as host and watermark images, selected for comparison with other methods.
3：List of Experimental Equipment and Materials:
MATLAB software is used for simulation. No specific hardware or materials are mentioned.
4：Experimental Procedures and Operational Workflow:
Steps include decomposing images with SFLCT (host in two levels with d=1, watermark in one level with d=2), applying SVD to details matrices, embedding watermark components with strength factors αA and αD, and using inverse SVD and SFLCT. Robustness is tested against various attacks like noise addition, filtering, compression, and geometric manipulations.
5：Data Analysis Methods:
Imperceptibility is measured using PSNR, WPSNR, and SSIM; robustness is measured using normalized correlation (NC). Comparisons are made with existing schemes.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
