A hybrid Fuzzy rule based Neuro-Genetic approach for fingerprint recognition system

Accurate fingerprint recognition system is the problem encountered in the field of biometric identification. Use of back propagation network for recognition problems resulted in inconsistent and unpredictable performance. The proposed approach using soft computing tools like Fuzzy, Neural and Genetic overcomes low recognition rate, low accuracy and increased time of recovery. The stages of this approach are Fingerprint Enhancement, Fuzzy based minutiae extraction and Neuro-Genetic based recognition. Fuzzy rule based systems are used for eliminating the false minutiae from the fingerprints. Optimization of neural network parameters is done using Genetic algorithm. This increases recognition accuracy and reduced training time. The system is trained and tested on FVC 2004 database.