| In this thesis, we develop and test a new image-adaptive watermarking system, "DIGI-COP", in the discrete cosine transform (DCT) domain, for copyright protection of digital images. "DIGital Image COpyright Protector" or short DIGI-COP generates a new type of watermark and embeds it in the image such that it becomes virtually invisible to the human eye. DIGI-COP'S performance is compared to the classical Block-DCT algorithm by Koch et al. and several improvements in terms of fidelity (imperceptibility) and security have been achieved.; In addition to the watermark extraction process, a novel Error Correction technique is developed to increase overall system accuracy and reliability against corruptly decoded bits. Extensive experimentations with over 100,000 encode/decode operations on thousand selected color images show DIGI-COP's high resilience to image processing tasks such as: JPEG image compression, several Gaussian noise levels, different image blurring kernels, two levels of image sharpening, six image scaling operations, color inversion and image cropping. |
