An efficient robust watermarking scheme for varying sized blocks

Abstract

Abstract:The idea of partitioning a host image into small nonoverlapping blocks and embedding the watermark in these blocks based upon some property of the blocks has been in use for many years and is still very popular. While most of the watermarking techniques use a constant strength factor for the whole image, in the present watermarking scheme a flexible strength factor is used, which is derived from the standard deviations of the discrete wavelet transform coefficients of the selected high entropy nonoverlapping blocks. The presented watermarking scheme explains the reverse relationship between imperceptibility and robustness by various plots involving adjusting the parameter of strength factor. A watermarked image of any desired peak signal to noise ratio can be prepared very easily by using this scheme.Three versions of the proposed watermarking scheme are presented here. In the first version the host image is partitioned into 8 × 8 nonoverlapping blocks and these blocks are used in the watermarking. Similarly, in the second and third versions, the host image is partitioned into 16 × 16 and 32 × 32 blocks respectively for their use in watermarking. We compare the results of these three versions with other watermarking schemes and illustrate the greater robustness of all versions of our watermarking scheme. The novelty of the scheme lies in the use of three sizes of nonoverlapping blocks and the flexible strength factor for every block of the image.