Forensic Handwriting and Signatures Using Machine Learning Techniques

Given the significance of the messages in these crimes and the signatures in the content of those messages, there is a growing need in the field of forensic analysis to verify these crimes and accusations, as well as to look for evidence in a variety of ways before reaching a final decision regardin...

Full description

Saved in:
Bibliographic Details
Published in2023 2nd International Engineering Conference on Electrical, Energy, and Artificial Intelligence (EICEEAI) pp. 1 - 6
Main Authors Alsarhan, Ayoub, Aljaidi, Mohammad, Samara, Ghassan, shdaifat, Andaleeb Al, Alhayajnah, Razan, Khafajeh, Hael Al, Qasem, Mais Haj, Alsarhan, Tamam
Format Conference Proceeding
LanguageEnglish
Published IEEE 27.12.2023
Subjects
Accuracy
Analytical models
criminal investigation
Fingerprint recognition
Forensics
Machine learning
Privacy
Shape
Online AccessGet full text

Cover

More Information
Summary:Given the significance of the messages in these crimes and the signatures in the content of those messages, there is a growing need in the field of forensic analysis to verify these crimes and accusations, as well as to look for evidence in a variety of ways before reaching a final decision regarding the accused individual. In this study, a thorough business model was created as evidence of tremendous relevance, mission, and strong privacy as a fingerprint that separates one individual from another. The suggested model will take into account a number of factors, including the accused's name and the location and date of the incident. To eliminate the need for investigators to guess due to unclear and suspicious handwritten messages containing evidence-either conclusive evidence or evidence that aids in the process of tracking events to reach conclusive evidence-a tool is developed to translate the messages into readable, understandable, and clear content. These resources are designed to help criminal investigators reach the best possible conclusions. The proposed machine was trained using a set of numbers, words, and signatures that had various shapes and patterns. This allowed the system to recognize and validate the signature in all of its forms, including some that were difficult to understand. The proposed model achieved results of more than 98%, compared to the previous studies in this area. (Abstract)
DOI:10.1109/EICEEAI60672.2023.10590549