Brigitte Wirtz

Stroke-based time warping for signature verification

This paper presents a new technique for dynamic signature verification. A dynamic programming (DP) approach is used for function-based signature verification. Dynamic data such as pressure is treated as a function of positional data and therefore evaluated locally. Verification is based on strokes as the structural units of the signature. This global knowledge is fed into the verification procedure. The application of a 3D non-linear correlation of the signature signals uses the stroke index as the third DP index. In conjunction with the definition of a finite state automaton on the set of reference strokes the system can handle different stroke numbers, missing or additional strokes correctly. The correct alignment of matching strokes is determined simultaneously to the signature verification process; an additional alignment stage before the actual nonlinear correlation is obsolete.

Average prototypes for stroke-based signature verification

A signature verification system basically consists of two units, the verification and the enrolment subsystem. Much work has been done developing verification techniques for dynamic signature verification whereas the influence of reference construction has not been extensively studied. The paper reports how the verification rates of a given system can be increased by ameliorating the enrolment technique. An efficient enrolment technique that performs a time- and position-based averaging of representative input signatures for a stroke-based verification approach is presented. The approach is capable of dealing with varying stroke structures, missing or additional strokes, and, in the presence of those, determines the most natural stroke structure of the reference in parallel with the construction of the average signature prototype. It is shown that, in comparison to the frequently used technique of selecting one signature out of an initial set of input signatures, the error rate especially of stroke-based approaches can be significantly reduced.

Parameter reduction and personalized parameter selection for automatic signature verification

This paper presents a signature verification system based on 84 personally selected parameters. An initial set of 300 parameters is first reduced globally by statistical methods. A personalized parameter selection is performed on the reduced parameter set by use of several norms and a linear classifier. The personalized parameter sets of different signers justify that parameter selection should be personalized. Verification results for personalized parameters are given for different norms and compared to results achieved with a function-based system. A method for combining both approaches and the results achieved with this extended system are presented as well.

3-D pose estimation by an improved Kohonen-Net

A primary task in the process of three-dimensional object recognition given a two-dimensional image lies in the model-based pose estimation in space. Given a known set of correspondences between three-dimensional model points or different model features and two-dimensional image points or image features: What are the transformation and model parameters which convert the three-dimensional model into the observed two-dimensional image.