Multi-image Feature Map-Based Watermarking Techniques Using Transformer

Nowadays, protecting multimedia data is a significant challenge because of the advancement technology and software. The embedding process heavily relies on watermarking to accomplish security in terms content authentication, proof ownership, tamper detection. Our objective develop an invariant watermark that can survive different signal-processing attacks. We presented unique hybrid technique (DWT-QR-SWT) multi-image features generated as using Transformer encoder-decoder model. encoded image are subsampled PCA order decrease dimensionality image. first two images used watermark1 next watermark2 produce multi-watermark feature maps. To embed watermark, DWT-QR decomposition has been applied original image1. On primary watermarked image, Level Stationary Wavelet Transform (SWT) were secondary watermark2. At extraction phase, tampered recovered by passing extracted input transformer decoder. A increases capabilities also achieves two-level detection, localization, recovery. With PSNR 59.05 dB, testing result demonstrates great resilience improved imperceptibility.