Handwriting analysis is a standard forensics practice to assess the identity of a person from written documents. Forensic document examiners consider different features related to the motion and pressure of the hand, as well as the shape of the different characters and the spatial relationship among them. While examiners rely on standard protocols, documents are generally processed manually. This requires a significant amount of time and may lead to a subjective analysis which is difficult to replicate. Automated forensics tools to perform handwriting analysis from scanned documents are desirable to help examiners extract information in a more objective and replicable way. To this aim, in this paper we present GRAPHJ, a forensics tool for handwriting analysis. The tool has been designed to implement the forensics protocol employed by the “Reparto Investigazioni Scientifiche” (RIS) of Carabinieri. GRAPHJ allows the examiner to (1) automatically detect text lines as well as the different words within the document; (2) search for a specific character and detect its occurrences in the handwritten text; (3) measure different quantities related to the detected elements (e.g., character height and width) and (4) generate a report containing measurements, statistics and all parameters used during the analysis. The generation of the report helps to improve the repeatability of the whole process. We also present a set of experiments to assess the compliance of GRAPHJ with respect to conventional handwriting analysis methods. Given a set of handwritten documents, the experiments compare measurements and statistics produced by GRAPHJ to those obtained by an expert forensics examiner performing classic manual analysis.
[1]
Eric J. W. Visser,et al.
Abramoff MD, Magalhaes PJ, Ram SJ. 2004. Image Processing with ImageJ. Biophotonics
,
2012
.
[2]
Roy Huber,et al.
Handwriting Identification: Facts and Fundamentals
,
1999
.
[3]
I. Evett,et al.
A Study of the Variation in the Dimensions of Genuine Signatures
,
1985
.
[4]
Sargur N. Srihari,et al.
On-Line and Off-Line Handwriting Recognition: A Comprehensive Survey
,
2000,
IEEE Trans. Pattern Anal. Mach. Intell..
[5]
Sebastiano Battiato,et al.
Multimedia Forensics: discovering the history of multimedia contents
,
2016,
CompSysTech.
[6]
Jan Seaman Kelly,et al.
Scientific Examination of Questioned Documents, Second Edition
,
2006
.