Fadable ink writing recognition based on laser-induced breakdown spectroscopy and machine learning
Yu Liu, Shen Li, Xiang Han, Cong Wang · Laser Physics Letters · 2025
Abstract A method for restoring fadable ink writings using laser-induced breakdown spectroscopy (LIBS) combined with machine learning was proposed. This research employed dimensionality reduction and clustering analysis to process LIBS data, significantly improving analytical efficiency and accuracy. Unlike traditional chemical detection methods, this approach minimizes chemical damage to samples while ensuring operator safety. Compared to physical restoration techniques, it achieves higher accuracy and lower operational costs. Experimental results demonstrated stable recognition performance in complex environments. For samples stored over extended periods, the method successfully restored handwriting. Repeated detection tests confirmed preserved clarity. The proposed method was validated through analysis of various ink samples, including those from the high-temperature fadable pen and sodium chloride (NaCl) solution, showcasing its effectiveness across different ink types. This work highlights a novel technique for recognizing fadable ink writings, with micro damage to samples and providing reliable analysis for samples stored over extended periods. This method avoids damage to samples from chemical reagents, is non-irritating to the human body, and is cost-effective. The technique can effectively restore fadable ink traces in important documents, including commercial contracts and legal evidence, providing an efficient solution for forensic document authentication, and offers a novel approach to information decryption in the field of encryption.