A simple image-adaptive watermarking algorithm with blind extraction

摘要

In this paper, we propose an adaptive, DCT based, blind image watermarking algorithm. The adaptability is developed, while exploring the properties of the human visual system (HVS), to embed robust watermarks in DCT matrices while making them the least visible in the space domain. In the algorithm, binary watermarks are embedded in image blocks having gray level variations sufficient to mask a certain level of signal alterations. The orientations of the variations in such a block are detected by examining the locations of the most significant DCT coefficients in the matrix and a particular DCT coefficient is then chosen for watermark embedding with a view to making the gray level changes resulting from the embedding have a similar orientation with the original image signal to mask the watermark better. Moreover, the embedding strength is made to depend on the DC component, as it indicates the brightness of the image block, to further reduce the visibility of the watermark. The algorithm is computationally simple because it extracts the needed features directly from the DCT coefficients. Simulation results show that the algorithm is highly robust against JPEG compression even at low quality factors while maintaining high visual quality. Furthermore, the computational simplicity of the algorithm makes it very useful in real-time applications.