Optimizing Imperceptibility and Robustness Using Selective DCT Coefficients for Image Copyright Protection
Majid Rahardi, Afrig Aminuddin, Ferian Fauzi Abdulloh, Nafiatun Sholihah, Ferda Ernawan · Engineering Technology & Applied Science Research · 2025
This paper presents an invisible watermarking technique for image copyright protection based on the Discrete Cosine Transform (DCT). The proposed method balances imperceptibility and robustness against various image manipulations, such as JPEG compression, filtering, and noise attacks. By embedding the watermark into high-frequency DCT coefficients, the technique ensures minimal visual distortion while preserving watermark integrity. The experimental results demonstrate that the proposed method achieves high Peak Signal-to-Noise Ratio (PSNR) values, averaging 51.81 dB across multiple test images, and excellent structural similarity with an average Structural Similarity Index Measure (SSIM) of 0.9993. The proposed method also shows strong resilience to image tampering, with Normalized Cross-Correlation (NC) values of 0.7593 and low Bit Error Rate (BER) values of 0.1190 under various attack scenarios. These results highlight the method's effectiveness in protecting image copyright while maintaining computational efficiency.