Achieving Higher Stability in Watermarking According to Image Complexity

One of the main objectives of all watermarking algorithms is to provide a secure method for detecting all or part of the watermark pattern in case of the usual attacks on a watermarked image. In this paper, a method is introduced that is suitable for any spatial domain watermarking algorithm, so that it can provide a measure for the level of robustness when a given watermark is supposed to be embedded in a known host image. In order to increase the robustness of the watermarked image, for a watermark of $M$ bits, it was embedded N=s\times M times, where $s$ is a small integer. Doing this, the entire image is divided into 16 equal size blocks. For each block, the complexity of the sub-image in that block is measured. The amount of repetition of the watermark bits saved in each block is determined, according to the complexity level of that block. The complexity of a sub-image is measured using its quad tree representation. This approach not only secures the watermarked image with respect to usual attacks, but also, enables one to save longer bit patterns of the watermark, while maintaining a good level of similarity between the original image and the watermarked one. For evaluating the performance of this method, it has been tested on 2000 images having low, medium and high levels of complexity and the result have been compared with the same set of images, without considering the complexity of sub-images in blocks. The new method provided 17% higher stability.