Nowadays digital contents are rapidly and inexpensively being distributed around the world by using the internet. This issue has become a source of great concern for multimedia content producers. Copyright protection of digital contents in virtual environments requires a reliable approach. Watermarking technology was introduced to solve this problem and now it is widely in use. The process of hiding owner-related information in digital contents was named watermarking. Since images are the most abundant form of digital contents, digital image's watermarking is the most common kind of watermarking. In this routine the owner's information may be embedded in spatial or transform domain of host images. Due to fidelity and strength requirements in noisy situations, recent watermarking methods favor transform domain watermarking. There is a trade-off between watermarking strength and fidelity. The use of more resistant watermarking jeopardizes the fidelity of the watermarked image. Because of that, presenting watermarking methods that satisfy these two features is one of the challenging issues in watermarking technology. In this thesis resistant and perspicuous watermarking methods are presented for digital gray level images. In these methods a binary logo that identifies content's owner will be concealed in Hadamard transform coefficients of host image blocks. The first proposed method employs a compression technique to embed watermark. In this method, watermark information is stored in regions with high texture and entropy. Region selection is done by use of the entropy measure of each image block. The owner identifier is obtained by informed watermark extraction. In this process the host image is used for determining watermarked blocks and watermark coefficients. In the second method the watermark logo is redundantly hidden in mid frequencies of Hadamard transform. To do so a number of matrices are created from the same-frequency coefficients of Hadamard transform. Then these matrices are decomposed into their bit planes. The watermark logo is replaced with the plane that provides more robustness and fidelity. In this method watermark detection needs no information from the host image and the watermark is blindly extracted. Watermark extraction routine is based on voting between different copies of watermark that were hidden in the host image. While these watermarking methods are robust against compression, crop, and filtering attacks, they also possess sufficient fidelity. Keywords: Bit plane, Compression, Hadamard Transform, Watermarking.
