<![CDATA[The increasing accessibility of digital image editing tools has made image forgery a growing concern in forensic, legal, and medical imaging contexts, where content authenticity is critical. This study proposes a fragile watermarking scheme based on non-overlapping 2 x 2 pixel block matrices integrated with the MD5 cryptographic hash function to enable precise tamper localization in digital images. Watermark embedding is restricted exclusively to the 1-bit Least Significant Bit (LSB) space, while the remaining 7-bit Most Significant Bit (MSB) of each block is processed using MD5, truncated to 4 bits, and embedded through a bitwise OR operation. The scheme was evaluated on 30 samples from the CoMoFoD forgery dataset through watermark embedding, spatial manipulation attack simulation, blind extraction, and localization, with performance measured using PSNR, SSIM, and detection accuracy. The results show an average PSNR of 51.09 dB and SSIM of 0.9977, indicating no perceptible visual degradation, alongside an average detection accuracy of 93.51%, confirming high sensitivity to spatial tampering with minimal false positives. These findings demonstrate that micro block-level hashing effectively balances imperceptibility and fragility, offering a computationally lightweight yet precise instrument for digital image forensics.]]>
Copyrights © 2026
