CNN-Based HVC Watermark Extraction for Print-and-Capture Scenarios

Fengshou Tao; Linghua Guo; Xiaofeng Zhao

doi:10.6919/ICJE.202609_12(9).0022

Authors

Fengshou Tao College of Bioresources Chemical and Material Engineering, Shaanxi University of Science and Technology, Xi’an 710021, China
Linghua Guo College of Bioresources Chemical and Material Engineering, Shaanxi University of Science and Technology, Xi’an 710021, China
Xiaofeng Zhao College of Bioresources Chemical and Material Engineering, Shaanxi University of Science and Technology, Xi’an 710021, China

DOI:

https://doi.org/10.6919/ICJE.202609_12(9).0022

Keywords:

Halftone Visual Cryptography; Print-and-capture; Convolutional Neural Network; Printed-media Anti-counterfeiting.

Abstract

Objective: This study addresses the difficulty of obtaining complete and clear digital extractions of halftone visual cryptography (HVC) watermarks under print-and-capture conditions. Methods: We constructed a differentiable print-and-capture degradation model that represented major physical degradation processes. Training samples were generated online by randomly combining degradation parameters. A convolutional neural network (CNNs) decoder was designed to recover 32 × 32 watermarks. The digitally stacked image was the primary input, while the registered watermarked image and digital key were auxiliary inputs. Multilevel downsampling established spatial correspondence between image features and watermark blocks. Multiscale dilated residual modules and channel attention extracted watermark features from degraded halftone dots. Results: The CNN decoder achieved a balanced correct decoding rate of 99.16% under simulated compound degradations. Its average decoding rate was 96.9% for real print-and-capture samples. This result exceeded direct digital stacking by 43.4 percentage points and blockwise XOR extraction by 13.8 percentage points. Conclusion: The CNN decoder learned the nonlinear mapping between degraded halftone-dot features and watermarks. It provides a practical approach for mobile HVC watermark extraction in print-and-capture scenarios.

Downloads

Download data is not yet available.

References

[1] Zou, K. C. (2021). Research on digital image watermarking technology [Master's thesis]. Xidian University. https://doi.org/10.27389/d.cnki.gxadu.2021.002305

[2] Wang, Y. Y. (2021). Research on image encryption and its security based on visual cryptography [Master's thesis]. Xi'an University of Technology. https://doi.org/10.27398/d.cnki.gxalu.2021.000475

[3] Guo, J. M., & Seshathiri, S. (2023). Visually encrypted watermarking for ordered-dithered clustered-dot halftones. IEEE Transactions on Circuits and Systems for Video Technology, 33(8), 4375–4387. https://doi.org/10.1109/TCSVT.2023.3237688

[4] Fu, M. S., & Au, O. C. (2002). Data hiding watermarking for halftone images. IEEE Transactions on Image Processing, 11(4), 477–484. https://doi.org/10.1109/TIP.2002.999680

[5] Guo, Y. F., Au, O. C., Wang, R., & et al. (2018). Halftone image watermarking by content aware double-sided embedding error diffusion. IEEE Transactions on Image Processing, 27(7), 3387–3402. https://doi.org/10.1109/TIP.2018.2815181

[6] Wang, Y. Y., Ni, R. R., Zhao, Y., & Xian, M. (2018). Watermark embedding for direct binary searched halftone images by adopting visual cryptography. Computers, Materials & Continua, 55(2), 255–265. https://doi.org/10.3970/cmc.2018.01732

[7] Rosen, J., & Javidi, B. (2001). Hidden images in halftone pictures. Applied Optics, 40(20), 3346–3351. https://doi.org/10.1364/AO.40.003346

[8] Basak, O., & Gaurav, S. (2009). Continuous phase-modulated halftones. IEEE Transactions on Image Processing, 18(12), 2718–2734. https://doi.org/10.1109/TIP.2009.2028367

[9] Guo, J.-M., & Seshathiri, S. (2019). Watermarking in dot-diffusion halftones using adaptive class-matrix and error diffusion. ECTI-CIT Transactions, 13(1), 1–8.

[10] Oztan, B., & Sharma, G. (2010). Multiplexed clustered-dot halftone watermarks using bi-directional phase modulation and detection. In 2010 IEEE International Conference on Image Processing (pp. 981–984). IEEE. https://doi.org/10.1109/ICIP.2010.5651620

[11] Oztan, B., & Sharma, G. (2010). Per-separation clustered-dot color halftone watermarks: Separation estimation based on spatial frequency content. Journal of Electronic Imaging, 19(4), Article 043007. https://doi.org/10.1117/1.3497615

[12] Ulichney, R., Pollard, S., & Gaubatz, M. (2014). Fast mobile stegatone detection using the frequency domain. In NIP & Digital Fabrication Conference (Vol. 30, No. 1, pp. 237–242). Society for Imaging Science and Technology. https://doi.org/10.2352/ISSN.2169-4451.2014.30.1.art00059_1

[13] Tancik, M., Mildenhall, B., & Ng, R. (2020). StegaStamp: Invisible hyperlinks in physical photographs. In 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 2114–2123). IEEE. https://doi.org/10.1109/CVPR42600.2020.00219

[14] Fang, H., Zhang, W. M., Zhou, H., & et al. (2019). Screen-shooting resilient watermarking. IEEE Transactions on Information Forensics and Security, 14(6), 1403–1418. https://doi.org/10.1109/TIFS.2018.2878541

[15] Qin, C., Li, X. M., Zhang, Z. Y., & et al. (2024). Print-camera resistant image watermarking with deep noise simulation and constrained learning. IEEE Transactions on Multimedia, 26, 2164–2177. https://doi.org/10.1109/TMM.2023.3293272

[16] Ma, Z. H., Fang, H., Yang, X., & et al. (2025). RoPaSS: Robust watermarking for partial screen-shooting scenarios. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 39, No. 18, pp. 19332–19339). https://doi.org/10.1609/aaai.v39i18.34128