Automatic Allograph Matching in Forensic Writer Identification

Abstract

A well-established task in forensic writer identification focuses on the comparison of prototypical character shapes (allographs) present in handwriting. In order for a computer to perform this task convincingly, it should yield results that are plausible and understandable to the human expert. Trajectory matching is a well-known method to compare two allographs. This paper assesses a promising technique for so-called humancongruous trajectory matching, called Dynamic Time Warping (DTW). In the first part of the paper, an experiment is described that shows that DTW yields results that correspond to the expectations of human users. Since DTW requires the dynamics of the handwritten trace, the "online" dynamic allograph trajectories need to be extracted from the "offline" scanned documents. In the second part of the paper, an automatic procedure to perform this task is described. Images were generated from a large online dataset that provides the true trajectories. This allows for a quantitative assessment of the trajectory extraction techniques rather than a qualitative discussion of a small number of examples. Our results show that DTW can significantly improve the results from trajectory extraction when compared to traditional techniques.

Dynamic time warping: A new method in the study of poor handwriting

“…They are also compared on the continuity criterion, which forces the matching between a point of the first curve and the nearest point of the second curve. If the relative temporal order of a point in curve 1 is about similar to the relative temporal order of a point of curve 2 (i.e., falls within a relative distance of plus or minus 0.13 -a value between 0 and 1 that was based on previous studies [14]), then these two points fulfill the continuity criterion (see Figure 1; for clarification and an exact definition of these criteria, see [14]). For each point of curve 1 all points of curve 2 that satisfy these requirements are evaluated and the point of curve 2 with the minimum distance to the point of curve 1 is chosen to form a pair of matching points.…”

“…DTW distances are in arbitrary units because all letters were first rescaled in size (between -.5 and +.5) to derive the individual prototype (IP), the child's "personal average", for each condition. In this study the prototype was calculated by the Merge-samples algorithm (for details see [14]), a procedure that creates a DTW distance from the IP for every single letter. Two extreme examples of the distribution of these DTW distances are presented in Figure 2.…”

“…This socalled trajectory-matching technique, originally designed by Kruskal and Liberman [9] for speech recognition purposes, was adapted by Niels, Vuurpijl, and Schomaker [14] for written character recognition and forensic document examination. With DTW, trajectories of pen movements are compared using both spatial and temporal information.…”

“…By objectively analyzing the spatial-temporal patterns, DTW captures the essential character of writing, i.e., the overall shape of its graphic output. The trajectory-matching technique devised by Niels et al [14] not only offers new perspectives for fundamental research comparing various models of handwriting, but also for the analysis of the distorted shapes produced by people with handwriting disorders.…”

Automatic writer identification framework for online handwritten documents using character prototypes

“…Such state-of-the-art algorithms and techniques make use of a template matching approach that assigns handwriting styles to prototype templates which are representative of the handwriting styles [9,[17][18][19][20]. The prototypes attempt to model the writing styles of the writers as close as possible, based on features extracted from the online handwritten documents.…”