Image orientation watermarking technique for copyright protection

• Main Author: Jasim, Hend Muslim
• Format: Thesis
• Language: English
• Published: 2016
• Online Access: http://psasir.upm.edu.my/id/eprint/69368/1/FSKTM%202016%2037%20-%20IR.pdf; http://psasir.upm.edu.my/id/eprint/69368/
• Institution: Universiti Putra Malaysia

Image copyright protection utilizes watermarking as one of the Digital Right Management techniques to protect digital images. Watermarking involves embedding a signature into a given image to produce a watermarked copy. This copy is meant to prevent any deliberate or unintentional attempt to remove the signature from its content. In return, a copyright gains its validity based on such robust characteristics. Watermarking attracts much interest in algorithm design to enable a signature to be embedded on secrecy and encryption beds. Watermarking imitates steganography in information hiding but with alternative priorities, it focuses on the cover and not on the hidden message. The use of watermarking has still not achieved a meaningful level of reliability in copyright protection applications as to be able to resolve infringement claims in the courts. This is due to two main obstacles. The first involves the need to consider original image quality as the main evidence over any other clues. The second obstacle on watermarking reliability in copyright protection involves the lack of an adequate common measure that proves the superiority of one technique over another and then hampers improvements to the efficiency and quality of algorithms To address this problem, this research proposes a new image-orientation watermarking technique overcoming the obstacles. The proposed technique intentionally corrupts the original images under owner signature control to generate copies for public use. Corruption is effected intrinsically by image operations such as resizing, which acts to relocate the original from its default position to an alternative over a presumed space. The design utilizes Principal Component Analysis and Blind Noise Level Estimation to resemble these transitions over image resizing operations. In addition, it formulates two measuring parameters, for evaluating protection techniques, namely Protection Requirement Parameter (PR) and Distance Decision Measure (D). The experiment has two phases which are Pre-Protection design and, Protection and Infringement Resolve. In the Pre-Protection design, it investigates the proofing on image operation corruption to support the principle used in showing the preliminary impact of data corruption made by resizing. In the second phase, using Protection and Infringement Resolve, it proceeds to cover the performance of the main two procedures of the protection technique along the validity parameters devised in this work. Results set a PR threshold of 0.05 to be totally convincing as an efficient requirement controlling parameter. Compared to the existing benchmarking noise measures of PSNR, MSE, MAXERR and L2RAT analysis of variance (ANOVA) and linearity are applied to support inferences from sample limits towards population boundaries in order to establish rigid validity on PR. The main conclusions of this research highlight the contradictions in the use of watermarking as a tool in image protection as the basis for proposing an alternative design. Digital watermarking techniques are usually complex and necessarily involve the need to satisfy many conflicting requirements and tradeoffs. Unlike these techniques, the proposed work adopts an image-oriented foundation that does not treat the owner’s signature as paramount but, instead, makes the customer’s image traceable through the owner’s signature on a transition path that uses original image quality as the main basis for copyright resolution. Results obtained affirm the validity of this design basis along with the proposed parameters that make comparison among techniques possible.