A Hybrid Watermarking Scheme Based on Arnold Cat Map Against Lossy JPEG Compression

Abstract

Watermarking is a data hiding method for multimedia elements such as image, video, audio, and software. Several works have been conducted in the spatial frequency domains for cover images. An essential demand in the watermarking algorithm is to be resistant to common attacks. In this work, the authors investigate the robustness and reliability of the transform domain watermarking against the Lossy JPEG compression, which is widely used in digital imaging applications and hardware. In addition, we propose a transform domain digital watermarking algorithm that resists the JPEG compression in low frequencies. The Discrete Fourier transform (DFT) and the Singular value decomposition (SVD) are selected as the embedding domain where the binary logo watermark is iteratively scrambled using the Arnold Cat Map (ACM) before embedding for robustness. Two main metrics are used for evaluation the scheme; Peak signal to noise ratio (PSNR) and Normalized Correlation (NC). Experimental results show very promising results with a PSNR value above 45 dB while the NC value remains above 0.9 even after high compression with a Quality Factor of only 1%.