Document signature is a powerful technique used to determine whether a message is tampered or valid. Recently, this concept was extended to optical codes: we demonstrated that the combined use of optical techniques and machine learning algorithms might be able to distinguish among different classes of samples. In the present work, we produce nano particle encoded optical codes with predetermined designs synthesized with a 3D printer. We used conventional polylactic acid filament filled with metallic powder to include the effect of nano-encoding for unique polarimetric signatures. We investigated an interesting real-world scenario, that is, we demonstrate how a single class of codes is distinguished among a group of samples to be rejected. This is a difficult unbalanced problem since the number of polarimetric signatures that characterize the true class is small compared to the complete dataset. Each sample is characterized by analyzing the polarization state of the emerging light. Using the one class-support vector machine algorithm we found high accuracy figures in the recognition of the true class codes. To the best of our knowledge, this is the first report on implementing optical codes with nano particle encoded materials using 3D printing technology.
