Traceable Image Data Sharing through Blockchain and Reversible Watermarking

This paper proposes a traceable image data sharing system that integrates reversible watermarking and blockchain technology. A 1024-bit signature derived from image content and the applicant data, is embedded into the image as the watermark. We integrate blockchain technology to record copyright information and auxiliary watermarking variables of the shared data to provide immutable and transparent ownership records. To strengthen security, Arnold transform is applied to scramble the watermark and embedded in the discrete cosine transform (DCT) coefficients by taking into account the contrast sensitivity function to maintain imperceptibility. In the event of unauthorized sharing or misuse of image data, the embedded watermark can be extracted and traced, allowing us to identify and track responsibility for unauthorized image use. Experiments were conducted using 10 color images from diverse domains, including general images (USC-SIPI), medical datasets (CHASEDB1, ISIC, KVASIR), and biometric datasets (FERET). Experimental results show that the method achieves high imperceptibility with average peak signal-to-noise ratio (PSNR) values of 56.28 dB, and average structural similarity index (SSIM) scores close to 1.0 across test payloads ranging from 1 Kb to 16 Kb, robustness against common distortions and adversarial attacks with watermark normalized correlation (NC) scores above 0.92, and accurate traceability across 1,000 distributed copies with various distortions and adversarial attacks. Comparative evaluations confirm that the proposed method outperforms existing watermarking schemes in both robustness and traceability performance.