Auth2X: Secure Biometric Authentication System

Tech Stack:

• Python
• C++
• MySQL
• Tkinter
• OpenCV
• SecuGen SDK
• Minutiae-Based Matching

---

🔐 Overview

Auth2X is a dual-mode biometric authentication system that supports:

• Face recognition (via OpenCV-based facial encoding)
• Fingerprint matching (via SecuGen Hamster Plus and custom minutiae comparison)

It supports encrypted storage of biometric data using Fernet AES encryption and integrates with a Tkinter GUI for both registration and authentication.

---

🧠 Key Features

• 🔐 Fingerprint & Face Authentication
• 🔑 Encrypted Biometric Data Storage
• 🧬 Minutiae Extraction and Matching (No third-party matcher)
• 🗃️ User info + biometric data stored in MySQL
• 🖥️ C++ Executable for Fingerprint Capture using SecuGen SDK
• 🧪 Match Ratio-based fingerprint comparison with logs

---

📚 Fingerprint Matching Method

This project uses a minutiae-based fingerprint matching pipeline that works as follows:

• Captured .dat image files (260x300) are loaded as raw binary
• Preprocessing includes binarization, skeletonization, and noise filtering
• Minutiae points (ridge endings & bifurcations) are extracted from the fingerprint skeleton
• Matching is performed by comparing geometrical proximity (Euclidean distance < 10 pixels) and type similarity
• Match ratio is calculated based on overlapping features
• If the match ratio > 0.65, authentication is accepted

All extracted templates are Fernet-encrypted before being stored in the database.

---

🧰 Libraries Used

• numpy – Image array manipulation
• mysql-connector-python – Database connection
• cryptography.fernet – AES encryption of biometric templates
• scikit-image – Image skeletonization and preprocessing
• tkinter – GUI interface for user interaction
• opencv-python – Face capture and recognition
• sv-ttk – Dark theme styling for Tkinter GUI

---

🗂️ Folder Structure

Auth2X/
├── gui/
│   ├── home_gui.py
│   ├── login_gui.py
│   ├── register_gui.py
│   └── result_gui.py
├── face_authentication/
│   └── face_auth.py
├── Face_registration/
│   └── face_registeration.py
├── fingerprint/
│   ├── capture/
│   ├── config/
│   │   ├── config.json
│   │   ├── db_config.json
│   │   └── secret.key
│   ├── store_template.py
│   ├── match_template.py
│   ├── match_utils.py
│   └── fingerprints/

---

🛠️ Setup Instructions

1. ✅ Prerequisites

• Python 3.10+
• MySQL Server
• Visual Studio 2019+
• SecuGen SDK (FDx Pro SDK)
• Python dependencies:

pip install mysql-connector-python cryptography opencv-python-headless numpy scikit-image sv-ttk pillow

---

2. 💾 MySQL Schema

CREATE DATABASE auth2x;
USE auth2x;

CREATE TABLE users (
    id INT AUTO_INCREMENT PRIMARY KEY,
    username VARCHAR(50) NOT NULL UNIQUE,
    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
);

CREATE TABLE biometric_data (
    id INT AUTO_INCREMENT PRIMARY KEY,
    user_id INT NOT NULL,
    biometric_type ENUM('face', 'fingerprint') NOT NULL,
    data BLOB NOT NULL,
    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    FOREIGN KEY (user_id) REFERENCES users(id)
);

---

3. 🔧 Configuration

config.json

{
  "encrypt_fingerprint": true
}

db_config.json

{
  "host": "localhost",
  "user": "root",
  "password": "yourpassword",
  "database": "auth2x"
}

secret.key

from cryptography.fernet import Fernet
with open("secret.key", "wb") as f:
    f.write(Fernet.generate_key())

---

🧱 C++ Fingerprint EXE Build

• Use Visual Studio to build CaptureFingerprint.cpp
• Link with sgfplib.lib
• Place all required DLLs next to the EXE
• set your fingerprints directory path in sln
• Output: CaptureFingerprint.exe in x64/Debug/

---

🔐 How to Run

python gui/home_gui.py

---

🔄 Workflow Overview

🧍‍♂️ Fingerprint Module

✅ Registration:

• Capture fingerprint via SecuGen SDK
• Store image as .dat
• Extract minutiae from thinned skeleton
• Encrypt (if enabled in config.json)
• Save in MySQL biometric_data table

🔐 Authentication:

• Capture new fingerprint
• Extract minutiae
• Fetch and decrypt stored template
• Compare using:
  • Minutiae type match (ending/bifurcation)
  • Euclidean distance < 10 px
• Match if ratio > 0.65
• Terminal logs:
  Matches: 24, Ratio: 0.75

👤 Fingerprint Contributors:

• Ramlah Munir – LinkedIn
• Talal – LinkedIn

---

🧑‍🦱 Face Module

✅ Registration:

• Capture live webcam image using OpenCV
• Detect face using face_recognition library
• Extract 128-dimension face encoding
• Encrypt the encoding with Fernet
• Store securely in MySQL

🔐 Authentication:

• Fetch encrypted encoding from DB
• Decrypt using Fernet key
• Compare live webcam face with stored encoding using compare_faces()
• Authenticate if match returns True

🖼️ Uses real-time webcam view with prompt to press s for scan.

👤 Face Recognition Contributors:

• Ayaan Ahmed Khan – LinkedIn
• Mohammad Umar Nasir – LinkedIn

---

🧪 Debugging & Logs

• Fingerprint .dat storage:
  • username.dat and username_live.dat
• Logs printed from matcher:
  • Matches: X, Ratio: 0.YYY
• Face module logs:
  • Decryption failures
  • Encoding mismatches

---

❓ FAQ

Q: Low match ratio?
A: Re-align finger, check capture quality, avoid moisture and blur.

Q: Face not detected?
A: Use good lighting, keep background simple, position face fully in frame.

---

📩 Contact

• Fingerprint: Ramlah's LinkedIn, Talal's LinkedIn
• Face: Ayaan's LinkedIn, Umar's LinkedIn