Visual attention model based dual watermarking for simultaneous image copyright protection and authentication

Abstract

In this paper, a visual attention model based dual watermarking scheme has been proposed for simultaneous image copyright protection and authentication. The suggested method simultaneously embeds two watermarks inside the host image. First of all, the robust watermark is embedded in the most visually salient region by Hessenberg factorization technique, which has proven to give better results in terms of imperceptibility and robustness. Subsequently, the second watermark acts as a fragile watermark for the purpose of tamper detection and localization. The fragile watermark is infused inside the image via fragile 8 Bit Plane Slicing, as we require for the watermark to be highly responsive to majority of geometric, and non-geometric impairments. The proposed scheme achieves an average Peak Signal to Noise Ratio and Structural SIMilarity Index of 46.88 dB and 0.9897 respectively. Experimental results depict that the projected scheme is extremely robust against most kinds of geometric, signal-processing, and hybrid attacks with an average Normalized Cross Correlation of 0.9401. Moreover, due to the incorporation of the fragile watermark, it can detect and localize tampering of an image with extreme accuracy.