Individuality of Handwriting

Abstract

Motivated by several rulings in United States courts concerning expert testimony in general and handwriting testimony in particular, we undertook a study to objectively validate the hypothesis that handwriting is individualistic. Handwriting samples of one thousand five hundred individuals, representative of the US population with respect to gender, age, ethnic groups, etc., were obtained. Analyzing differences in handwriting was done by using computer algorithms for extracting features from scanned images of handwriting. Attributes characteristic of the handwriting were obtained, e.g., line s e p aration, slant, character shapes, etc. These attributes, which are a subset of attributes used by expert document examiners, were used to quantitatively establish individuality by using machine learning approaches. Using global attributes of handwriting and very few characters in the writing, the ability to determine the writer with a high degree of confidence was established. The work is a step towards providing scientific support for admitting handwriting evidence in court. The mathematical approach and the resulting software also have the promise of aiding the expert document examiner.

Gender Classification by Means of Online Uppercase Handwriting: A Text-Dependent Allographic Approach

“…Actually, the following groups are considered: gender, handedness, age (under 24/above 45) and origin (US-born/non-US-born). The features considered are the micro-features described in [35], and the samples used for training are taken from the well-known CEDAR letter database, also described in [35]. When it comes to gender classification, the results are quite surprising since some individual characters show an unexpectedly high performance: letter ''b'' attains a 70 % performance, digit ''3'' a 67 %, letters ''m'' and ''y'' a 66 % and letters ''Y'' and ''a'' a 65 %.…”

Automatic writer identification from text line images

“…Among the well-known text-dependent approaches, the extraction of macro and micro features is one of the most comprehensive [2]. Macro features can be extracted at document, paragraph, line, and word or character level and consist of pen pressure, writing movement, stroke information, slant, and height features.…”

“…It could be argued that all document examiner features could eventually be computational features-when the correct algorithms have been defined. Although, computational algorithms allow defining many features which are impossible to be measured by human, it is fact that most of the document examiner features are not yet computable [2]. The main difficulties to compute the document examiner features are that they are text dependent and their automatic extraction would require recognition, moreover some of the features such as embellishment, legibility or writing quality is qualitative.…”

“…As the main factors affecting the constitution of a person's handwriting, such as biological and cultural, differ for each person, handwriting is a distinguishing biometric like voice. Writer individuality rests on the hypothesis that each individual has a consistent handwriting which is distinct from the handwriting of another individual [2]. Hence, the individuality of handwriting makes the identification of the writer of a handwritten sample possible.…”

“…Most of these techniques depend on distance measures [3,16,[22][23][24][25]. K-Nearest Neighbor (K-NN) classifier and neural networks are other widely used methods [1,2,12,[24][25][26]. There are also seldom used methods such as Bayes classifier [17,26], Gaussian Mixture Model (GMM) [19], Hidden Markov Model (HMM) [18] and Support Vector Machines (SVM) [15].…”