pyfnprint fingerprint recognition library

Disclaimer

This is a simple fingerprint recognition library written in Python3. The library implements various fingerprint preprocessing and recognition methods along with evaluation and database splitting utilities. The library is focused on classical image-processing-based approach and does not use more advanced machine learning techniques such as neural networks.

The library is created out of purely academic and research-related goals and does no focus on efficiency and/or is not created to be used in production.

References

(Almost) all references that have been used to write the library are located in the references directory

TLDR

Prepare environment (assuming ${PWD} is the root of the repository)

python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install -r requirements.txt
export PYTHONPATH=$PYTHONPATH:$PWD

Split db files into a random train/test parts

python3 scripts/dbsplit.py -f 75 -i 75 db/PNG data

Enroll train data

python3 example/example.py -e data data/templates

Identify test data

python3 example/example.py -p data data/templates

Evaluate the predictions

python3 scripts/evaluation.py data/test.csv data/prediction.csv

Directory tree structure

src/fplib - the main library sources

• image.py - wrapper type for a fingerprint image with a filename formatted like {id}_{number}.{extension} and lazy reading
• preprocess.py - preprocessing with various image quality enhancing functions and functions for extraction of ridge characteristics
• binarize.py - binarization and various opencv wrappers
• filter.py - filters with support of custom kernels and gabor filtering
• minutae.py - extraction of minutae points and core point detection
• feature.py - feature extraction and comparison
• plot.py - plotting module for various stages of preprocessing

db/PNG - toy database

• A tiny database that contains 128 high quality fingerprint images - 8 for each user. Images are labeled as {id}_{number}.png where id is the identifier of a finger and number is the identifier of a specific fingerprint image of the finger
• More challenging database can be found for example at the FVC competition

src - command line utilities

• scripts/dbsplit.py - сommand line tool for splitting any amount of fingerprint images into a single filesystem database with train/test structure
• scripts/evaluation.py - сommand line tool for evaluating predictions

example

• example/example.py - an example program that uses fplib sources to create a fingerprint identification model

Original image	Skeletonized
Original image	Segmented
Ridge orientations	Minutae