This is repository containing implementation of experiments mentioned in the paper concerning offline verification of handwritten signatures. Original paper (Croatian) link: https://drive.google.com/file/d/1akPpLKbRkJUbThOnsDSyMiHOgVIyXicJ/view?usp=sharing
1st chapter training set: https://www.kaggle.com/divyanshrai/handwritten-signatures
2nd chapter training set: http://www.iapr-tc11.org/mediawiki/index.php?title=ICDAR_2009_Signature_Verification_Competition_(SigComp2009)
Image preprocessing functions are implemented in
preprocessor.py script. In order to perform whole preprocessing as described in referenced
paper, just run preprocess() method.
Feature extraction methods are implemented in feature_extractor.py
script. In order to retrieve all features from preprocessed image, just
run extract() method.
Euclidian model is implemeted via euclid_model class in
euclidian_model.py script. All the other experiments from the first
chapter are conducted in simple_models.py script.
Experiments from the second chapter using Neural Networks and SVM are
implemented in complex_models.py script.