Advanced Anticounterfeiting: Angle-Dependent Structural Color-Based CuO/ZnO Nanopatterns with Deep Neural Network Supervised Learning

Abstract

Current anticounterfeiting technologies rely on deterministic processes that are easily replicable, require specialized devices for authentication, and involve complex manufacturing, resulting in high costs and limited scalability. This study presents a low-cost, mass-producible structural color-based anticounterfeiting pattern and a simple algorithm for discrimination. Nanopatterns aligned with the direction of incident light were fabricated by electrospinning, while CuO and ZnO were grown independently through a solution process. CuO acts as a reflective layer, imparting an angle-dependent color dependence, while ZnO allows the structural color to be tuned by controlling the hydrothermal synthesis time. The inherent randomness of electrospinning enables the creation of unclonable patterns, providing a robust anticounterfeiting solution. The fabricated CuO/ZnO nanopatterns exhibit strong angular color dependence and are capable of encoding high-density information. It uses deep learning algorithms to achieve an average discrimination accuracy of 94%, with a streamlined computational structure based on shape and color features to achieve a processing speed of 80 ms per sample. The training images are acquired with standard high-resolution cameras, ensuring accessibility and practicality. This approach offers an efficient and scalable next-generation solution for anticounterfeiting applications, including documents, currency, and brand labels.