An Invisible, Robust and Secure DWT-SVD Based Digital Image Watermarking Technique with Improved Noise Immunity

摘要

The chances of copyright violation and privacy have been increased due to growth of networking and technology. Digital watermark is useful to verify the integrity of the content and for authenticity of an owner. A digital watermark is a digital signal or pattern inserted into a digital document (text, graphics and multimedia presentations). Watermarking is one of the promising solutions for ownership authentication, tamper detection and protection of digital content. The watermark inserted into the image should be invisible so that it may not be forged by clever attackers; reversible so that at the receiving end there is exact recovery of the watermark and the host image. However, the exact recovery of the watermark is not possible when the image is attacked by noise, cropped, rotated, median filtered, low or high pass filtered and many more such alterations or attacks. Digital images are commonly pre-processed by JPEG before being sent through internet. Attacks cause degradation of the image and it may result in alterations in the pixel values of the image. Such alterations in the image data are harmful for detection of watermark payload and host image. Encryption is used for adding the security to the watermark image embedded to the host image. Arnold transform is used for encrypting the watermark image. The watermark is embedded into the host image in the frequency domain. DWT is used to convert the host image into the frequency domain. Singular values of the 8×8 blocks LL band coefficients are calculated. This further adds to the robustness and embedding capacity of the watermarking algorithm.