Development of DWT–SVD based Digital Image Watermarking for Multi-level Decomposition

Abstract

In the age of digital data, the safety of digital content has gradually become paramount important. The demand for digital data in various forms such as text, images, video and audio has increased manifold. There are lots of vulnerabilities concerning digital data. It can easily be attacked, forged and manipulated to produce illegal copies. In the watermarking process, information, i.e., watermark image is inserted into a host data and extracted later in case of Copyright violation; owner of the content can demonstrate his ownership by regaining the embedded watermark. A watermark must be capable of being recovered even if a piece of content is modified or changed by several attacks. This paper uses frequency-domain technique, and the multi-level decomposition of the host image is done by using the combination of two-dimensional DWT (2D DWT) and SVD techniques. Here, the host/cover image is decomposed at six different levels separately, and the watermark image is inserted in the region of lowest frequency. After inserting a watermark, the host/cover image is reconstructed using an inverse DWT technique. By exposing the watermarked image to different image processing attacks such as pepper and salt, speckle and rotation, the robustness of various levels has been analyzed. The embedded watermark is extracted afterward and matched with the original one based on PSNR and MSE. The comparative experimental results show that the decomposition of the host image at the third level is found to be more suitable for watermarking among all other levels.