Colloquium Marius Bulacu

Dit jaar gaat CoVer ook de (afstudeer-)colloquia promoten. Zodoende kunnen leden een duidelijker beeld krijgen van het onderzoek aan het eigen instituut. Wil je meer van de colloquia weten? Kijk dan op www.ai.rug.nl/nl/colloquia

Deze week een stafcolloquium: Marius Bulacu - \"Computer Analysis of Handwriting Individuality\"
Automatic person identification using scanned images of handwriting is an interesting pattern recognition problem with direct applicability in the forensic field. Interest for this area has increased in the scientific community after 9/11 and the anthrax letters. Approaching this problem raises a number of important research themes in computer vision:

How can individual handwriting style be characterized?
What representations are most appropriate?
What performance can be achieved?
We developed new and very effective techniques for automatic writer identification that use probability distribution functions (PDFs) extracted the handwriting images to characterize writer individuality.

Our methods operate at two levels of analysis: the texture level and the character-shape (allograph) level.

At the texture level, we use contour-based directional PDFs that encode orientation and curvature information to give an intimate characterization of the individual handwriting style.

In our analysis at the allograph level, the writer is considered to be characterized by a stochastic pattern generator of ink-trace fragments, or graphemes. The PDF of these simple shapes in a given handwriting sample is characteristic for the writer and is computed using a common codebook of graphemes obtained by clustering.

The representations or features used to encode individual handwriting style must be designed to be independent of the textual content of the handwritten sample. In our approaches the computer is completely unaware of what has been written in the sample. The handwriting is merely seen as a texture characterized by some directional probability distributions or as a simple stochastic shape-emission process characterized by a grapheme occurrence probability.

The development of our writer identification techniques takes place at a time when many biometric methods (e.g. iris, fingerprint, face) undergo a transition from research to real full-scale deployment. Our methods also have practical feasibility and hold the promise of direct applicability.

A html-base demo of our writer identification system will also be presented in the colloquium.