A feature adaptive image watermarking framework based on Phase Congruency and Symmetric Key Cryptography

In this paper, a Phase Congruency based digital color image watermarking algorithm is proposed which provides a higher degree of robustness against attacks and excellent imperceptibility. Here, Phase Congruency has been used to detect the local feature regions of the host image and then the watermark has been infused into it using a new technique called ‘Adaptive α-β Blending’. An accurate Human Visual System modeling has been incorporated via Lifting Wavelet Transform to take the full advantage of perceptual watermarking. The coefficients of the α-β blending are selected adaptively based on the Phase Congruency feature map of the host image. Furthermore, the watermark is secured with a cryptographic algorithm called Arnold’s Cat Map to prohibit further eavesdropping. From rigorous testing, results indicate that our approach is robust against various geometric, non-geometric and combined attacks while maintaining a sublime imperceptibility.