Secure Image Provenance: Blockchain-Based Solution for Image Forgery and Similarity

Document Type



The proliferation of digital images has become ubiquitous in today’s interconnected world. However, with the rise of sophisticated image manipulation tools, the number of manipulated images has skyrocketed, putting the trust and reputation of individuals and industries at risk. Addressing this issue necessitates effective methods to detect manipulated images, find similar images, and track their provenance for better trust. In this work, we propose a comprehensive approach to address this problem, which consists of three main components: image manipulation detection, image similarity detection, and using blockchain technology for image provenance.

In the first component, we develop an image forgery localization method using contrastive learning to differentiate between manipulated and untampered regions. This approach focuses on generalization across various manipulations, providing enhanced detection capabilities. For the second component, we propose a deep learning-based method for learning fine-grained image similarity directly from images, which can be applied to find similar images from a blockchain database. Finally, in the third component, we employ blockchain technology to create a secure, transparent, and immutable record of image provenance. By integrating the International Standard Content Code (ISCC) with our image manipulation and similarity detection models, we can reliably track image provenance and enhance trust in digital images.

In summary, this work presents a comprehensive approach to detecting manipulated images, identifying similar images, and tracking image provenance using blockchain technology. Our proposed image manipulation detection and image similarity detection models have shown promising results, providing a reliable solution for various industries. By integrating these models with a blockchain-based image provenance system, we can enhance trust in digital images and ensure their integrity in an increasingly interconnected world.

First Page


Last Page


Publication Date



Thesis submitted to the Deanship of Graduate and Postdoctoral Studies

In partial fulfillment of the requirements for the M.Sc degree in Machine Learning

Advisors: Dr. Abdulmotaleb Elsaddik, Dr. Karthik Nandakumar

Online access available for MBZUAI patrons