TY  - JOUR
AU  - Aksu, Emine
AU  - Karakoyun, Murat
PY  - 2026
DA  - 2026/01/31
TI  - DCT-SVD Based Hybrid Approach for Digital Watermarking of Medical Images
JO  - ICCK Transactions on Swarm and Evolutionary Learning
T2  - ICCK Transactions on Swarm and Evolutionary Learning
JF  - ICCK Transactions on Swarm and Evolutionary Learning
VL  - 2
IS  - 1
SP  - 1
EP  - 18
DO  - 10.62762/TSEL.2025.219908
UR  - https://www.icck.org/article/abs/TSEL.2025.219908
KW  - digital watermarking
KW  - DCT
KW  - SVD
KW  - medical images
KW  - patient privacy
KW  - image processing
AB  - This study addresses invisible watermarking techniques aimed at preserving patient privacy during the sharing of medical images. Digital watermarking is a significant method for protecting the confidentiality of patient data by securely embedding personal information into medical images. In this study, three different strategies were developed and compared using a DCT-SVD-based hybrid invisible watermarking technique. In the first method, the host image and the watermark were of the same size, and direct embedding was applied. In the second method, the host image was divided into sixteen 128x128 blocks, and the watermark was segmented accordingly and embedded into each block individually. In the third and proposed method, non-diagnostic regions of the image—referred to as dead zones—were automatically detected, and the watermark was embedded only into these areas. This approach preserved the relevant medical data while minimizing image distortion. When the scaling factor was set to 0.01, PSNR values exceeded 40 for most images, and SSIM values were above 0.9. The results demonstrated that the proposed method outperformed the other two in terms of both imperceptibility and robustness.
SN  - 3069-2962
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Aksu2026DCTSVD,
  author = {Emine Aksu and Murat Karakoyun},
  title = {DCT-SVD Based Hybrid Approach for Digital Watermarking of Medical Images},
  journal = {ICCK Transactions on Swarm and Evolutionary Learning},
  year = {2026},
  volume = {2},
  number = {1},
  pages = {1-18},
  doi = {10.62762/TSEL.2025.219908},
  url = {https://www.icck.org/article/abs/TSEL.2025.219908},
  abstract = {This study addresses invisible watermarking techniques aimed at preserving patient privacy during the sharing of medical images. Digital watermarking is a significant method for protecting the confidentiality of patient data by securely embedding personal information into medical images. In this study, three different strategies were developed and compared using a DCT-SVD-based hybrid invisible watermarking technique. In the first method, the host image and the watermark were of the same size, and direct embedding was applied. In the second method, the host image was divided into sixteen 128x128 blocks, and the watermark was segmented accordingly and embedded into each block individually. In the third and proposed method, non-diagnostic regions of the image—referred to as dead zones—were automatically detected, and the watermark was embedded only into these areas. This approach preserved the relevant medical data while minimizing image distortion. When the scaling factor was set to 0.01, PSNR values exceeded 40 for most images, and SSIM values were above 0.9. The results demonstrated that the proposed method outperformed the other two in terms of both imperceptibility and robustness.},
  keywords = {digital watermarking, DCT, SVD, medical images, patient privacy, image processing},
  issn = {3069-2962},
  publisher = {Institute of Central Computation and Knowledge}
}