Improved implementation of digital watermarking techniques

Abstract

With the advancement of high speed computer networks, multimedia storage and transmission technology, anenormous amount of information is being communicated in digital form. There is a constant threat to copyright, ownership and integrity of digital data. Therefore information security has become an emerging area of research. Security of images is an area of major concern as digital images are available all over World Wide Web. Numerous watermarking techniques have been developed to protect images from illegal manipulations. In this project, spatial domain and transform domain have been proposed. Histogram shifting falls under the spatial domain method and spread spectrum falls under transform domain method. For the proposed Histogram Shifting, modification of the histogram and shifting is applied for embedding the stage, add to that our contribution in this method incorporated the threshold concept to improve the visual quality of the host image. In the proposed Spread Spectrum, a chaos-based spread spectrum watermarking algorithm is developed in the DCT domain for still image. The most significant feature of chaos is its sensitivity to initial conditions. This characteristic of chaos has been used successfully for secure watermarking applications. Local properties of the image and the features of the human visual system are considered in order to optimize the watermark strength in addition to the incorporated luminance masking effect of the HVS in the masking image, since the human eye is less sensitive to change in regions with high brightness as well as in very dark regions in comparison to mid-grey regions, where the distortion is most noticeable. The contribution only uses luminance masking for enhancing the visual quality of images in the embedding stage instead of using a combination of different masks and that leads to less complexity. The performance of the proposed watermarking schemes have been evaluated by using the watermarked images of size 512×512, and the watermark (payload) is of the same size as the host image in the spread spectrum with a different amount of variables "random bit streams" used in the histogram sniffing. The simulations are performed in MATLAB 7 software environment. The PSNR metric and SSIM with GMSD are applied to measure the degradation of the images. A comparison is made between the results of the proposed algorithms with each other and also compared with the best method namely Bit Plane Mapping and demonstrates that the proposed methods are superior in performance and especially the proposed Spread Spectrum.