Digital watermarking is important for copyright protection, but challenges remain in maintaining a balance between imperceptibility and robustness. This study evaluated Discrete Cosine Transform (DCT)-based watermarking on JPEG images by inserting binary watermarks at the intermediate frequency coefficient of the DCT block 8×8. The method was tested against JPEG compression, Gaussian noise, Gaussian blur, and Salt & Pepper noise to assess the visual quality and durability of the watermark. The results show high visual quality with an average PSNR of 39.99 dB and a SSIM of 0.96, so the watermark is almost imperceptible. Robustness testing yielded a BER of 1.5% on Gaussian noise, 5.2% on JPEG compression, and 8.2% on Salt & Pepper noise. These findings prove that the DCT method is able to maintain image quality and maintain a watermark on most interferences, despite degrading performance in impulse attacks and aggressive compression. In practical terms, this approach can be applied to digital image protection systems and forms the basis for the development of more robust hybrid or adaptive methods in the future.
Copyrights © 2025