Biometric Security Engines — Windows Hello & OpenCV Recognition

Overview

Biometric authentication is the cornerstone of offline security systems. I developed a suite of highly secure, offline-first biometric authentication applications:

1. Biometric Enabled Voting Machine (BEVM): An encrypted voting station prototype incorporating Windows Hello Fingerprint API to verify voter identities.
2. Face Recognition Attendance System: A real-time computer vision checkpoint using OpenCV to track check-ins and check-outs of candidates.

These projects were engineered to run strictly on local edge devices, prioritizing privacy and ensuring zero external network dependencies.

Technical Valuation: Replacement Cost: ₹6.8 Lakhs | Edge Cryptographic Engineering. Strict voter privacy hashing models.

---

Technical Features & Security Guardrails

1. Biometric Hardware Integration (winsdk/winrt)

The voting engine hooks directly into the local host's biometric hardware via the Windows Hello API (winsdk / winrt). The application locks the user interface until a successful biometric scan is registered. If hardware is missing, it falls back to a secure cryptographic simulator mode.

2. Edgewise Computer Vision Pipeline (OpenCV)

The attendance engine runs a real-time webcam video stream, detecting faces, calculating spatial face encodings, and comparing them against pre-registered student profiles.

• Data Persistence: Student face metadata is serialized into a local JSON store, allowing sub-150ms matching latency on standard CPU hardware.

3. Cryptographic Guardrails (AES-256)

To secure voter selections and candidate identities:

• Anonymity: Voter IDs are immediately hashed using SHA-256 prior to storage. No raw identities are ever written to the database.
• Vote Encryption: Vote choices are encrypted in the local SQLite database using AES-256 (Fernet). Decryption keys are stored in a separate keyfile (secret.key) and read only in memory for dashboard generation.