Elsevier

Forensic Science International

Volume 242, September 2014, Pages 266-273
Forensic Science International

Bayes factor for investigative assessment of selected handwriting features

https://doi.org/10.1016/j.forsciint.2014.07.012Get rights and content

Abstract

This paper extends previous research [1] on the use of multivariate continuous data in comparative handwriting examinations, notably for gender classification. A database has been constructed by analyzing the contour shape of loop characters of type a and d by means of Fourier analysis, which allows characters to be described in a global way by a set of variables (e.g., Fourier descriptors). Sample handwritings were collected from right- and left-handed female and male writers. The results reported in this paper provide further arguments in support of the view that investigative settings in forensic science represent an area of application for which the Bayesian approach offers a logical framework. In particular, the Bayes factor is computed for settings that focus on inference of gender and handedness of the author of an incriminated handwritten text. An emphasis is placed on comparing the efficiency for investigative purposes of characters a and d.

Introduction

Forensic examination of handwriting involves the description of handwriting features along with the study of their range of variation. On an analytical account, several studies [2], [3] have proposed a procedure based on Fourier analysis that allows scientists to describe the contour shape of loops of characters through a set of variables (so-called Fourier descriptors) and to study both the within-writers and the between-writers variability. Frameworks for evidence evaluation have also been developed and applied to cases under investigation. In particular, the use of the Bayes factor for assessing the value of evidence is now well known and regularly put forward in forensic literature [4].

The Bayes factor (Section 2) represents a coherent metric for evidence assessment in general (i.e., when a suspected writer is available for comparison purposes), but it can also be developed for investigative purposes [1]. The latter case involves settings in which no immediate suspect is available for comparison purposes so that examinations of handwriting concentrate primarily on characterizing general writer features (such as gender, or handedness). The main objective of this paper is to study the performances of a Bayes factor for assessing the value of handwriting features for such kind of investigative proceedings. In fact, anonymous handwriting regularly arises in contexts where, at least initially, no suspect is available. Thus, there will be no possibility in this setting for evaluating characteristics observed in a questioned document and those in reference (or control) material as it would be the case in a conventional evaluative scenario. Notwithstanding, the ‘recovered’ data, that is measurements made solely on a questioned document, can be valuable for informing ongoing investigations. For example, scientists may support investigative authorities in discriminating between more general hypotheses such as ‘the author of the questioned document (say, a threatening letter) is a male (or a female)’ or ‘the author of the questioned document is (or is not) a left-handed writer’. Information on such general writing features can help to reduce the pool of potential authors.

The paper is organised in theoretical and practical parts. Section 2 briefly defines the Bayes factor, in forensic science also generally referred to as the likelihood ratio. Here, the definitional differences between these two statistics will be clarified. Material, data and relevant models are illustrated in Section 3. The performances of the proposed approach for investigative purposes are presented and discussed in Section 4. Section 5 concludes the paper.

Section snippets

Bayes factor

The Bayes factor (BF) is given by the ratio between the posterior odds and the prior odds and measures the change produced by evidence in the odds when going from the prior to the posterior distribution in favor of one hypothesis as opposed to another.

In both evaluative and investigative settings, when the competing hypotheses are simple (the term ‘simple’ means that there is only one possible value for the null and the alternative hypothesis), the Bayes factor reduces to the likelihood ratio (

Material, data and models

Loops of characters a and d were extracted on documents written by 80 writers (students) from the Institut de police scientifique (University of Lausanne), aged from 20 to 22, who produced a total of, respectively, 7457 and 5216 observations of letter a and d.

Letters a and d were chosen following conclusions made in Ref. [3] where it has been shown that loops of a and d showed more characterising features regarding shape, allowing for better efficiency in the discrimination between the writers.

Gender classification

Various studies focused on handwriting features that could be used to help drawing an inference about the sex of a writer [8], [9], but apparently no feature or combination of features seemed to be peculiar to a specific gender. Recent research [1] focused on Fourier analyses on loops of character d.

Here, the study is extended to character a and the performances of the proposed approach for handwriting analysis expressed in terms of correct classification rates obtained with these different

Conclusion

This paper extends previous research [13], [1] where the contour shape of loop characters of type d – by means of Fourier analysis – has been studied for evaluative and investigative purposes. A new character, the letter a, has been studied and its performance in investigative settings (e.g., gender and handedness classifications) was compared to results obtained from character d. Results support findings from previous studies in evaluative settings [13]: character d offers better results than

References (14)

There are more references available in the full text version of this article.

Cited by (11)

  • Bayesian evaluation of dynamic signatures in operational conditions

    2022, Forensic Science International
    Citation Excerpt :

    Much progress was made through the meticulous description of methodology, as well as the validation of expert performance. This was achieved through population studies, as well as the application of computational techniques to strengthen pattern-matching examination [13,18–39]. Reliability claims were supported by statistical analyses related to mock cases built from large datasets containing known source writing [40–44].

  • Non-invasive optical micro-identification of ink verification in pen ink handwriting

    2020, Results in Chemistry
    Citation Excerpt :

    Therefore, determining which of the ink lines was made last [11,12] may help in solving the dispute as a recurrent matter in many police investigations and in courts. Generally speaking, handwriting with ink as primary method to calligraph and affix one's signature is widely used in document writing [13,14], document approval [15], contract signing and other fields [16–29], especially in the clear responsibility of signature, which plays an important role in distinguishing identity. The analysis of ink handwriting detection has important significance for technical division to determine economic responsibility and/or authority responsibility relationship.

  • Dynamic signatures: A review of dynamic feature variation and forensic methodology

    2018, Forensic Science International
    Citation Excerpt :

    Their approach is similar to Kupferschmid’s, but they additionally measure calibration and performance of the system. Gaborini et al. [210] apply and adapt the method of Marquis et al., based on Fourier Descriptors [71,211–214], and propose a multi-variate evidence evaluation approach. As can be seen through the literature review in this article, dynamic signature examination has gained the attention of the forensic examiner community, which is actively working to fill methodological gaps and develop robust and reliable methods.

  • Assessment of signature handwriting evidence via score-based likelihood ratio based on comparative measurement of relevant dynamic features

    2018, Forensic Science International
    Citation Excerpt :

    In order to evaluate the authorship based on the similarities and differences observed on questioned and reference handwriting, previous studies on automatically extracted features [3–5] have already contributed to help FHEs to quantitatively measure the features of handwriting and assess the value of handwriting evidence. The application of a likelihood ratio framework for handwriting evidence evaluation received particular attention [6–10]. In this paper, we will build on this framework.

  • Towards a Bayesian evaluation of features in questioned handwritten signatures

    2017, Science and Justice
    Citation Excerpt :

    However, there are a number of attempts in literature detailing individual aspects. As an example, a very recent study [34] related to Marquis' Fourier descriptors is available. Such study will serve us as a basis to integrate them into a more general framework.

  • Extraction and analysis of the width, gray scale and radian in Chinese signature handwriting

    2015, Forensic Science International
    Citation Excerpt :

    Various studies have shown that trained forensic document examiners perform significantly better than lay people in the analysis, comparison and identification processes based on handwriting evidence [5]. Much research on the applications of quantitative extraction methods for handwriting features [6–10] and on the statistics methods for forensic handwriting examination assessment [11–16] has been conducted to reduce the subjective aspect in forensic handwriting examination. Various features are extracted and applied for Chinese handwriting verification [17].

View all citing articles on Scopus
View full text