A secure anti-counterfeiting system based on perovskite microstructures

Counterfeiting is a growing global problem, with counterfeit goods, such as medicines and high-tech products, introduced into many markets, damaging brand reputation, reducing profitability, and posing serious health risks. Thus, this paper describes a novel anti-counterfeiting system that generates proof of a product’s authenticity by growing a novel perovskite-based microstructure. The created perovskite images (254) may be verified by the user using their mobile phone in both offline and online modes, with accuracies of 96% and 98%, respectively. The offline verification yielded a false acceptance rate (FAR) of 3% and a false rejection rate (FRR) of 7%, whereas no false acceptances or false rejections were observed for the online verification in the evaluated dataset. For the generation of such proofs, a GhostNet convolutional neural network is employed to generate embeddings of the images. Fuzzy storage and encryption are used to protect the genuine patterns and the exchanged information. Although more comprehensive user trials involving different smartphone models and uncontrolled acquisition conditions remain part of our future work, the results demonstrate the feasibility of distinguishing forged microstructures from genuine microstructure families under the evaluated conditions.