Artificial inteligence based classification techinques in signature verification

OData support
Supervisor:
Dr. Kővári Bence András
Department of Automation and Applied Informatics

The automatic check systems are more and more important these days, since they are more efficient and reliable. The automatic signature-verification one of these systems and it is one of the oldest systems of all. Scientist have been researching this system for decades even though there is no perfect solution for this challenge yet.

There are 2 different automatic signiture check systems at the moment. One of them is the online automatic signature-verification system where special and expensive tools are applied to record signatures. The other one is the off-line signature-verification system where the image of the signature is recorded. I have focused on the second one in my thesis.

While doing research I have analysed the use or 2 different applications. Both Matlab and Tensorflow are able to check signatures. Both Matlab and Tensorflow enable us to apply record systems based on artificial intelligence. I was going to try these solutions while achieving the highest possible accuracy.

This thesis describes classification based on Matlab and Tensorflow and compares efficiency in case of different input databases. There are 4O signatures from 2O signatories each and 15O features have been checked altogether.

Downloads

Please sign in to download the files of this thesis.