Authentication of Medical Staff with Protective Gear-Wearing: Utilization of Handwritten Letter Characteristics and Machine Learning

In recent years, various human identification technologies, such as facial recognition, have been increasingly adopted for security purposes. However, these conventional biometric systems face significant challenges in scenarios where individuals are required to wear protective clothing, which can obscure facial features and fingerprints. This paper introduces an innovative approach designed to overcome such obstacles by focusing on the individuality of handwritten characters, offering a viable alternative when traditional biometric identifiers are unusable. Our study proposes and evaluates a method that uses machine learning to analyze and learn the unique features of handwritten characters. This approach is independent of typical biometric traits such as facial features and fingerprints, thus providing a novel solution for identity verification in specialized environments like laboratories or hazardous material handling areas where protective gear is mandatory. We developed a model using a random forest algorithm trained on binary images of distinct handwritten characters written by participants. The training process involved extracting and learning the subtle differences in handwriting styles, strokes, and patterns that are difficult to replicate or disguise. The effectiveness of this methodology was validated through rigorous testing. Impressively, the trained model achieved a correct identification rate of 97.8%, underscoring the potential of handwriting-based identification as a robust and reliable security measure.