Off-Line Signature Verification Based on Fuzzy Modeling with the Optimal Number of Rules

This work presents an innovative approach for off-line signature verification based on fuzzy modeling. For feature extraction, both static and pseudodynamic features are extracted. By considering these features as fuzzy sets, a fuzzy model of r-rules for verification is constructed. In this model, the new membership functions are devised and the optimal number of rules is selected by using K-fold cross-validation which gives a reliable estimate of the verification performance. Finally, the experimental results confirm the effectiveness of the optimized technique when compared with the existing schemes.