yubarajshrestha / fingerprint-classification-system

Biometric recognition system using fingerprint images for Verification and Authentication - Universidad Nacional de San Agustin (Arequipa - 2017).

Fingerprint-Classification-System-Through-Modules

Fingerprint classification system using fingerprint orientantion feature vectors obtained after passinf through different modules - Universidad Nacional de San Agustin (Arequipa - 2017).

Modules

Data Storage Module

• NIST4 database

Data Preprocessing Module

• Preprocessing stages:
  • Fingerprint Histogram Equalization
  • Fingerprint Gabor Enhancement
  • Fingerprint Threshold Binarization
  • Fingerprint Thinning
• Final results:
  • NIST4 4000 thinned fingerprint images

Data Feature Extraction Module

• Two features extracted:
  • Region of Interes extraction through ANN detection algorithm
  • Orientation Map 100 and 400 features extraction algorithm
• Final results:
  • Manually extracted training database for roi block detection.
  • NIST4 4000 roi fingerprint images of 200x200px
  • 2 dat files containing: * 4000 NIST4 features vectors of 400 values from roi orientation map. * 4000 NIST4 features vectors of 100 values from roi orientation map.

Data Processing and Results Gathering( Classification and Results Module )

• Fingerprint classification system:
  • Fingerprint classification through Stacked Sparse Autoencoder using Keras.
  • Fingerprint classification models with different parameters.
• Final results:
  • Results from experiments of classification system
  • Nice classification model of 88% accuracy found during experimentation available and ready to be loaded in keras.
  • Spanish scratch Paper.