WATERMARKING is the process of inserting hidden information in an image by introducing modifications to its pixels with minimum perceptual disturbance. Arecent survey of major techniques, a fixed number of highest magnitude DCT coefficients are randomly perturbed, so that the watermark is placed to the perceptually significant components of the image. Even though the method is quite robust against signal manipulations, the original image must be present for watermark recovery. Recently, the pursuit of a scheme that doesn’t need the original image during watermark recovery has become a topic of intense research. This is partly due to practical issues, like the fact that the recovery process can be simplified without comparison with the original image. Also, in many instances, release of original material for any purposes is not desired or prohibited. Piva et al. developed a DCT-based scheme where the watermark can be identified by calculating the correlation between the watermark sequence and the DCT coefficients of the watermarked image. A similar scheme was proposed by Dugad applied to the DWT domain. In these methods, a significant correlation can be obtained only through a big number of coefficients (typically more than 10 000 ). Fridrich developed a method where a binary-valued watermark is inserted to the low-frequency region based on a mapping function and a spread spectrum signal is added to the midfrequency region.
A DCT-based image watermarking algorithm is described, where the original image is not required for watermark recovery, and is achieved by inserting the watermark in subimages obtained through subsampling.