An Efficient Approach for Acquisition of Fake Images using Blockchain and Machine Learning

Image tampering has become an effortless practice with recent advancements in image editing software with high-performance hardware. Fake images can affect any person's integrity or damage an organization's reputation. Hence, images must be processed in large quantities to provide a traceable and verifiable data set for further deep learning to detect tampered features in the images. The Blockchain works as a secure storage for the images and stores these images inside the blocks. These images can be accessed using the authentication rules defined by the administrator. Stored images are further verified and audited through smart contracts deployed on Blockchain. Blockchain can provide not only the image management issues between different objects but also save time and effort in processing the images. This paper covers a systematic approach for the acquisition of images spread across social media using an implementation of Blockchain and significant challenges for blockchain technology in the fake images domain. The implementation outcome reveals that the blockchain method significantly improves the storage, verification, and tracking of images using smart contracts. The authors modeled the image traceability problem as the interaction between the developed smart contracts such as Register Contract, Authentication Contract, and Verifier Contract and image tampering detection setup has been created for this model using a blockchain platform which is both quick and highly scalable. Experimental results validate that all procedures, such as blockchain configuration, smart contracts creation, authentication, tracing, and auditing of image transactions, have been done accurately as per the problem.