Biometric Template

Biometric Template -- Mathematical Representation of Biometric Data

A biometric templatebiometric templateBiometricMathematical representation of biometric data stored on card.Click to view → is a compact mathematical representation of a person's biometric features -- fingerprint minutiae points, iris texture codes, or facial geometry landmarks -- stored on a smart card for identity verification. Unlike a raw biometric image (which can be several hundred kilobytes), a template is typically 300 to 1,000 bytes, making it suitable for storage in the constrained EEPROM or flash memory of a smart card chip.

Template Extraction and Storage

During biometric enrollment, a sensor captures the raw biometric sample -- a fingerprint image, iris scan, or facial photograph. Feature extraction algorithms process the raw data to identify distinctive characteristics: minutiae points (ridge endings and bifurcations) for fingerprints, the IrisCode binary pattern for iris recognitioniris recognitionBiometricEye pattern recognition used in ePassportePassportApplicationPassport with embedded contactless chip.Click to view → and ID smart cards.Click to view →, or facial landmark distances and ratios. The resulting template is encoded in a standardized format (ISO/IEC 19794 series or proprietary) and written to the card during electrical personalization.

Template Security and Privacy

A critical property of biometric templates is their one-way nature: the template cannot be reverse-engineered to reconstruct the original biometric image, providing a measure of privacy protection even if the card is compromised. For additional security, templates may be encrypted before storage using AES keys managed by the card's secure element. The CBEFF standard (ISO/IEC 19785) wraps templates in a structured record format that includes metadata about the biometric type, quality score, and capture device, enabling interoperability across different systems and readers.

Matching and Accuracy

When verification is performed via Match-On-Card, the card's processor compares the candidate template against the stored reference using a matching algorithm. The algorithm computes a similarity score, and the match decision is based on a configurable threshold that balances FAR/FRR trade-offs. Smart cards used in ePassport and eID programs typically store multiple template types -- fingerprint plus facial image -- to support both automated border gates and manual identity checks. PIV cards store two fingerprint templates for the primary and secondary fingers.