Spectral Data Analysis for Forgery Detection in Official Documents: A Network-Based Approach

Abdulbasit Ali Al-Ameri M., Ciylan B., Mahmood B.

ELECTRONICS (Basel), vol.11, no.23, pp.1-17, 2022 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 11 Issue: 23
  • Publication Date: 2022
  • Doi Number: 10.3390/electronics11234036
  • Journal Name: ELECTRONICS (Basel)
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aerospace Database, Communication Abstracts, INSPEC, Metadex, Directory of Open Access Journals, Civil Engineering Abstracts
  • Page Numbers: pp.1-17
  • Keywords: digital forensics, forgery detection, unsupervised clustering, LIBS
  • Gazi University Affiliated: Yes


Despite the huge advances in digital communications in the last decade, physical documents are still the most common media for information transfer, especially in the official context. However, the readily available document processing devices and techniques (printers, scanners, etc.) facilitate the illegal manipulation or imitation of original documents by forgers. Therefore, verification of the authenticity and detection of forgery is of paramount importance to all agencies receiving printed documents. We suggest an unsupervised forgery detection framework that can distinguish whether a document is forged based on the spectroscopy of the document’s ink. The spectra of the tested documents inks (original and questioned) were obtained using laser-induced breakdown spectroscopy (LIBS) technology. Then, a correlation matrix of the spectra was calculated for both the original and questioned documents together, which were then transformed into an adjacency matrix aiming at converting it into a weighted network under the concept of graph theory. Clustering algorithms were then applied to the network to determine the number of clusters. The proposed approach was tested under a variety of scenarios and different types of printers (e.g., inkjet, laser, and photocopiers) as well as different kinds of papers. The findings show that the proposed approach provided a high rate of accuracy in identifying forged documents and a high detection speed. It also provides a visual output that is easily interpretable to the non-expert, which provides great flexibility for real-world application.