Fuzzified Deep Learning based Forgery Detection of Signatures in the Healthcare Mission Records

Ishu Priya,Nisha Chaurasia,Ashutosh Kumar Singh,Nakul Mehta,Abhishek Singh Kilak,Ahmed Alkhayyat

DOI: https://doi.org/10.1145/3641818

IF: 1.471

2024-01-24

ACM Transactions on Asian and Low-Resource Language Information Processing

Abstract:In an era subjected to digital solutions, handwritten signatures continue playing a crucial role in identity verification and document authentication. These signatures, a form of bio-metric verification, are unique to every individual, serving as a primitive method for confirming identity and ensuring security of an individual. Signatures, apart from being a means of personal authentication, are often considered a cornerstone in the validation of critical documents and processes, especially within the healthcare sector. In healthcare missions, particularly in the regions that are underdeveloped, hand-written records persist as the primary mode of documentation. The credibility of these handwritten documents hinges on the authenticity of the accompanying signatures, making signature verification a paramount safeguard for the integrity and security of medical information. Nonetheless, traditional offline methods of signature identification can be time-consuming and inefficient, particularly while dealing with a massive volume of documents. This arises the evident need for automated signature verification systems. Our research introduces an innovative signature verification system which synthesizes the strengths of fuzzy logic and CNN (Convolutional Neural Networks) to deliver precise and efficient signature verification. Leveraging the capabilities of Fuzzy Logic for feature representation and CNNs for discriminative learning, our proposed hybrid model offers a compelling solution. Through rigorous training, spanning a mere 28 epochs, our hybrid model exhibits remarkable performance by attaining a training accuracy of 91.29% and a test accuracy of 88.47%, underscoring its robust generalization capacity. In an era of evolving security requirements and the persistent relevance of handwritten signatures, our research links the disparity between tradition and modernity.

computer science, artificial intelligence

What problem does this paper attempt to address?

The paper attempts to address the issue of forged handwritten signature detection in medical task records. Specifically, the research team proposes a hybrid model combining fuzzy logic and convolutional neural networks (CNN) to verify the authenticity of handwritten signatures, ensuring the security and integrity of medical documents.

In the medical field, handwritten signatures remain an important means of identity verification, especially in underdeveloped areas. However, traditional offline signature recognition methods are inefficient, particularly when handling a large number of documents. Therefore, this study introduces an innovative signature verification system that uses fuzzy logic for feature representation and CNN for discriminative learning, providing an efficient and accurate solution. After 28 epochs of training, the hybrid model achieved an accuracy of 91.29% on the training set and 88.47% on the test set, demonstrating strong generalization capabilities.

Additionally, the study emphasizes the importance of enhancing security, preventing fraud, and optimizing workflow in medical environments. By combining the advantages of fuzzy logic and CNN, the system can address issues such as low-quality signature images, difficulty in signature cropping, and high similarity between forged and genuine signatures, thereby improving the security and reliability of medical documents.