Handwriting and Signature Authentication: Deep Learning Approaches to Questioned Document Examination

Book Title: Computational Criminology: AI Applications in Forensic Science and Criminal Justice

Editors: Dr. Xavier Louis, Dr. Surbhi Girdhar, Ms. Aswathi Chandran Nair, Mr. Ravi Kumar, and Ms. Nandini Katare

Chapter: 13

DOI: https://doi.org/10.59646/704/13

Author: Pallavi A Rao

Abstract

Questioned document examination is among the oldest applied forensic disciplines. Albert Osborn’s Questioned Documents, first published in 1910 and revised throughout the early twentieth century, codified the principles of handwriting and signature comparison that would define the discipline for nearly a century.[1] Those principles rested on two empirical premises: that the handwriting of an individual displays consistent identifying characteristics that distinguish it from the handwriting of others, and that a trained examiner can reliably evaluate the similarity between questioned and known specimens.[2] Both premises have been subject to sustained methodological scrutiny over the past three decades scrutiny that has prompted the discipline’s engagement with quantitative methods, computational similarity scoring, and, most recently, deep convolutional and recurrent neural networks. The methodological terrain of contemporary questioned document examination is divided into two related but distinct enterprises. Writer identification (sometimes called writer attribution) asks: given a questioned handwriting sample, which member of a population of candidate writers produced it? Signature verification asks: given a questioned signature, is it a genuine signature of the named individual, or is it a forgery? The two tasks share methodological infrastructure feature extraction, similarity scoring, classifier training but differ in their evidentiary contexts, error-cost structures, and the appropriate use of automated tools.[3]


[1]Albert S Osborn, Questioned Documents (2nd edn, Boyd Printing, 1929) 18.

[2]Roy A Huber and A M Headrick, Handwriting Identification: Facts and Fundamentals (CRC Press, 1999) 27.

[3]Sinha, A. K. (2026). Whistleblower Protection and Criminal Investigations in Corporations. Minnesota Journal of Business Law and Entrepreneurship (1) 452, 455.