Content area
Full text
Abstract - Since the Forensic Document Examiners (FDE) use inscriptions and different markings as clues to identify a writer, this paper presents an approach to writer identification based on graphometry. The framework proposed works at the level of ratio that is the relationship between two or more objects, in this case two or more components of handwriting (relative placement habits and word heights). The information provided by the feature extraction is used in a SVM (Support Vector Machine) as classifier. The experimental results show an identification rate equal to 80% considering all-against-all with 20 writers demonstrating comparable results in the literature.
Keywords: Graphometry; Writer identification; Feature Extraction; Classification: Forensic letter.
1 Introduction
In dispute cases, questions related to the writer identification in documents presented as evidence can be discussed in Court. The problem becomes greater when dealing with handwriting documents, since the attempts to fraud are more easily accessible, once high technology is not necessary to do them. In most cases, a writer and a pen are sufficient to accomplish the fraud.
According to Morris [1], the forensic handwriting identification is part of criminology and this analyses provide a great number of elements that affect a person's writing. This important area also know the relevance of systems of writing (paper, pen, arms, fingers and hand) and how they influence the writer since his childhood even his graphical mature writing.
Currently, the forensic handwriting identification is performed by experts using optical device and/or chemicals methods. According to Sheikholeslami et al. [2] and Fernandez-de-Sevilla et al. [3] the manual features extraction process is tedious and can provide doubts about the writer identification. In addition, different graphologists can extract the same features from a particular document in a different way. Then, the use of semi-automatic identification systems can be useful and helpfully to the experts. In this context, different approaches have been presented in research such as [4-13],
This work propose a framework to writer identification which applies a feature set at the level of ratios or relative relationships such as, relative placement habits and relative relationship between individual word heights. These features are based on graphometry and are used by the experts during their analyses.
This paper is organized as follows. Section 2 summarizes the...