Perceptual improvement of robust DCT-domain watermarking through wavelets based image fusion

In this paper, to improve perceptual quality of a robust discrete cosine transform (DCT)-domain watermarking scheme, a wavelet based image fusion is proposed instead of spatial masking techniques. The robust watermarking is achieved by inserting watermark sequences in the low frequency DCT coefficients of a host image. Then, the host and the watermarked image are decomposed using a two-level discrete wavelets transform (DWT). The image fusion is done by means of replacing the approximate and detail coefficients of the watermarked image with the new coefficients that obtained by applying a certain rule. Simulations with common attacks, such as JPEG compression, filtering, rescaling and noise addition, confirmed that the proposed method can significantly improve the perceptual quality of a watermarked image while maintaining the robustness.