In this dissertation, the methods of protection and authentication for digital images are presented. In the study of fragile image authentication, the method that can thwart the counterfeit attack is proposed in Chapter 2, and we analyze this method in order to prove the effect. In the study of semi-fragile image authentication, two image authentication methods with digital signature-based and digital watermark-based are presented in Chapter 3 and Chapter 4 respectively. These two methods can improve the disadvantages of related works, and the main contributions are: (1) can adjust the fragile degree by assigning the least authenticable bound of image and (2) the related theorems about the proposed methods are analyzed completely. Finally, a new issue and solution about semi-fragile image authentication are presented in Chapter 5. The main contributions of this method are: (1) only the spatial domain is adopted during feature generation and verification, making domain transformation process is unnecessary, (2) more reasonable non-malicious manipulated images (JPEG, JPEG2000 compressed and scaled images) than related studies can be authenticated, achieving a good trade-off of image authentication between fragile and robust under practical image processing, and (3) non-malicious manipulation is clearly defined to meet closely the requirements of sending them over the Internet or storing images.

The advances in recording, editing and broadcasting multimedia content in digital form motivate the protection of digital information against illegal use, manipulation and distribution. This thesis work focuses on one aspect of digital rights management (DRM), namely digital watermarking. Specifically, we study its use in copy protection, tamper detection and information hiding. We introduce three application-specific digital watermarking techniques. The first two algorithms, based on embedding film grain like noise and signal dependent watermarks, respectively, are designed for authentication applications. The advantage is that they are able to detect malicious tampering while being robust against content-preserving processes such as compression, filtering and additive noise. The third method, a reversible watermarking technique, is designed so that sensitive personal information can be embedded in medical images. Simulation results show that our proposed method outperforms other approaches in the available literature in terms of image quality and computational complexity.

The advances in recording, editing and broadcasting multimedia content in digital form motivate the protection of digital information against illegal use, manipulation and distribution. This thesis work focuses on one aspect of digital rights management (DRM), namely digital watermarking. Specifically, we study its use in copy protection, tamper detection and information hiding. We introduce three application-specific digital watermarking techniques. The first two algorithms, based on embedding film grain like noise and signal dependent watermarks, respectively, are designed for authentication applications. The advantage is that they are able to detect malicious tampering while being robust against content-preserving processes such as compression, filtering and additive noise. The third method, a reversible watermarking technique, is designed so that sensitive personal information can be embedded in medical images. Simulation results show that our proposed method outperforms other approaches in the available literature in terms of image quality and computational complexity.

To securely communicate information between parties or locations is not an easy task considering the possible attacks or unintentional changes that can occur during communication. Encryption is often used to protect secret information from unauthorised access. Encryption, however, is not inconspicuous and the observable exchange of encrypted information between two parties can provide a potential attacker with information on the sender and receiver(s). The presence of encrypted information can also entice a potential attacker to launch an attack on the secure communication. This dissertation investigates and discusses the use of image steganography, a technology for hiding information in other information, to facilitate secure communication. Secure communication is divided into three categories: self-communication, one-to-one communication and one-to-many communication, depending on the number of receivers. In this dissertation, applications that make use of image steganography are implemented for each of the secure communication categories. For self-communication, image steganography is used to hide one-time passwords (OTPs) in images that are stored on a mobile device. For one-to-one communication, a decryptor program that forms part of an encryption protocol is embedded in an image using image steganography and for one-to-many communication, a secret message is divided into pieces and different pieces are embedded in different images. The image steganography applications for each of the secure communication categories are discussed along with the advantages and disadvantages that the applications have over more conventional secure communication technologies. An additional image steganography application is proposed that determines whether information is modified during communication. Copyright / Dissertation (MSc)--University of Pretoria, 2012. / Computer Science / unrestricted

The great advance of communication technology comes with a rapid increase of disinformation in many kinds and shapes; manipulated images are one of the primary examples of disinformation that can affect many users. Such activity can severely impact public behavior, attitude, and belief or sway the viewers' perception in any malicious or benign direction. Additionally, adversarial attacks targeting deep learning models pose a severe risk to computer vision applications. This dissertation explores ways of detecting and resisting manipulated or adversarial attack images. The first contribution evaluates perceptual hashing (pHash) algorithms for detecting image manipulation on social media platforms like Facebook and Twitter. The study demonstrates the differences in image processing between the two platforms and proposes a new approach to find the optimal detection threshold for each algorithm. The next contribution develops a new pHash authentication to detect fake imagery on social media networks, using a self-supervised learning framework and contrastive loss. In addition, a fake image sample generator is developed to cover three major image manipulating operations (copy-move, splicing, removal). The proposed authentication technique outperforms the state-of-the-art pHash methods. The third contribution addresses the challenges of adversarial attacks to deep learning models. A new adversarial-aware deep learning system is proposed using a classical machine learning model as the secondary verification system to complement the primary deep learning model in image classification. The proposed approach outperforms current state-of-the-art adversarial defense systems. Finally, the fourth contribution fuses big data from Extra-Military resources to support military decision-making. The study proposes a workflow, reviews data availability, security, privacy, and integrity challenges, and suggests solutions. A demonstration of the proposed image authentication is introduced to prevent wrong decisions and increase integrity. Overall, the dissertation provides practical solutions for detecting manipulated and adversarial attack images and integrates our proposed solutions in supporting military decision-making workflow.