The scope of application of biometrics technology is expanding in the areas of identity verification and
technique of comparing characters written in two documents to determine whether they were created by the
same person. The handwriting verification expert determines the similarity of the marking, the pressure, and
the geometric features in a comprehensive manner. However, it is not possible to trust verification results
because the basis of each criterion is different and ambiguous to define. Therefore, the verification result as
evidence is not acceptable in court. To solve the problems, we use geometric features such as bifuaction,
ending, and corner points which were extracted by some digital image processing such as binarization,
component labeling, size filtering, and thinning in order to compare two handwritings. The final features are
selected by eliminating adjacent points. In order to normalize different image sizes, the coordinate values of
the feature points are multiplied by the magnification factor. Comparisons between detected feature point
sets are performed through measuring the modified Hausdorff distance. By analyzing the performance
through genuine and imposter matching, we confirmed the feasibility of objective handwriting verification.