Unobtrusive Watermarking for Copyright Preservation and Authenticity Verification in Digital Images Using Hybrid HVS-Based Technique

This research presents an innovative approach to hybrid photo watermarking designed to protect copyright and ensure authenticity with minimal detectable changes. Drawing inspiration from the human visual system, our method enhances copyright protection. In our proposed technique, watermark embedding and extraction involve examining the U components of the LF sub-band (LL) after Singular Value Decomposition (SVD) transformation across various threshold settings. Initially, specific image blocks undergo single-level Discrete Wavelet Transform (DWT) to produce four sub-bands per block, facilitating watermark extraction. The efficacy of our hybrid picture watermarking method, rooted in the HVS model, is assessed through tests involving median filtering, sharpening, JPEG compression, cropping, and Gaussian noise, all confirming its imperceptibility. Evaluation metrics such as PSNR assess the quality of watermarked images, while the BCR measures the quality of the extracted watermark.