Blind image watermarking via psychovisual-based relative modulation in DCT domain with performance optimized by GWO

A psychovisual-based relative modulation (PRM) scheme with performance fine-tuned by the grey wolf optimizer (GWO) is introduced to carry out effective blind image watermarking. This watermarking scheme starts with partitioning the host image into blocks of equal size and then applies the two-dimensional discrete cosine transform (DCT) to each block. Binary embedding in each DCT block is achieved by manipulating a strategically selected DCT coefficient with the embedding strength adaptively governed by a psychovisual model. The most appropriate DCT coefficient for PRM watermarking has been theoretically investigated and experimentally verified. The GWO plays the role of pursuing the optimal parameters that jointly attain a balanced performance in imperceptibility and robustness. Experimental results indicate that the proposed PRM-GWO scheme is competent to cope with a wide variety of attacks including image compression (JPEG and JPEG2000), noise contamination (white Gaussian, salt-and-pepper, speckle), filtering (lowpass, unsharp, median, Wiener), histogram equalization, geometric correction, and luminance adjustment. Furthermore, the PRM-GWO outperforms the other four DCT-based schemes in terms of the bit error rates of the recovered watermarks.