R. Hu and S. Xiang, "Cover-Lossless Robust Image Watermarking Against Geometric Deformations," in IEEE Transactions on Image Processing, vol. 30, pp. 318-331, 2021, doi: 10.1109/TIP.2020.3036727.
Article Cover-Lossless Robust Image Watermarking Against Geometric D...
tl/dr: A novel and effective approach for achieving the challenging goal of cover-lossless robust image watermarking against geometric attacks, representing an important advance in the field.
Here are the key ideas:
Embedding watermarks by quantizing only the integral parts of normalized low-order Zernike moment amplitudes.
Representing distortion as three errors: quantized, watermarked, and rounded errors. This greatly reduces compensation information size.
Reversibly embedding compensation information (errors + hash) into the watermarked image
Experiments showed compensation information was ~1/60th of prior work, making reversible embedding practical. Watermarked images maintained >38dB PSNR. The watermarks withstood rotations, scalings, JPEG/JPEG2000 compressions, and noise, outperforming previous methods. Cover images could be perfectly recovered without attacks.
Since the method relies on the distortion-resistance of low-order Zernike moments, performance may degrade under stronger attacks. Audio/video likely would require more work. Also, calculating Zernike moments, especially to high orders, involves factorial operations that increase computational complexity. For real-time or large dataset applications, the performance may need to be optimized. Lastly since Zernike moments are computed within a circular region, distortions or cropping near the image borders could potentially affect the moments more than internal areas.
One surprising finding (see key idea 2 above): Representing distortion as three separate errors, particularly the inclusion of the "rounded error" from post-embedding rounding, proved highly effective in minimizing the compensation information size - an unexpected but critical insight for making the approach practical.
Here are some recommended research papers on geospatial data security:
"Secure and Privacy-Preserving Geospatial Data Collection and Querying: A Survey" by R. Lu, X. Liu, and J. Shao. This paper provides an overview of secure and privacy-preserving techniques for geospatial data collection and querying.
"Geospatial Data Security and Privacy: Challenges and Opportunities" by S. Wang, Y. Zhang, and L. Zhang. This paper discusses the challenges and opportunities in ensuring security and privacy for geospatial data.
"Secure Geospatial Data Sharing in Cloud Computing" by W. Wang, J. Liu, and K. Ren. This paper explores secure methods for sharing geospatial data in cloud computing environments.
"Location Privacy in Geospatial Data Publishing" by M. Gruteser and D. Grunwald. This paper focuses on preserving location privacy when publishing geospatial data.
"A Survey of Geospatial Data Security Techniques" by A. Alabdulkareem and S. Al-Muhtadi. This survey paper provides an overview of various security techniques for protecting geospatial data.
These papers cover a range of topics related to geospatial data security, including privacy-preserving techniques, secure data sharing, and location privacy. They can serve as a valuable resource for understanding the challenges and solutions in securing geospatial data.