Ga Tech Cs 8903 Masters Project – Handwritten Signature Verification Discriminating Stroke Features Experiments

This document describes the progress made to date on the discriminating stroke features design. The approach for this filter in the system was to determine which of the strokes are most discriminating and then use those strokes to generate new words. The idea was that if a letter is composed of strokes, then by knowing which strokes are most discriminating, we will also know which letters are therefore most discriminating. From this, we can generate new words at random, composed of these letters. Presumably, it would be more difficult to forge a word generated at random than a signature because knowledge of the whole writing style for a particular writer would be needed to be able to forge new words on the fly.