Biometric Matching Systems

The integrity of a civil registry is achieved by biometric identifiers that correspond to a person uniquely - fingerprint, face, iris, voice and palm print. Based on proprietary smart indexing, our highly accurate and efficient biometric matching algorithms dramatically reduce the hardware requirements of large-scale ID projects.

AFIS

Decatur’s Automated Fingerprint Identification System is compliant with National Institute of Standards and Technology (NIST) standards. The speed and accuracy of its proprietary fingerprint matching technique makes it suitable for very large scale de-duplications.

Ranked amongst top 3 in MINEX test by NIST
Good performance at FVC-onGoing test
Above 99.9% accuracy for two fingers matching according to PFT II test by NIST
Low template sizes
Excellent image enhancement and segmentation capability

Face Recognition

With a detection accuracy of nearly 100%, Decatur’s proprietary face matching technique is used for both verification (1:1) and identification (1:N). Its adjustable learning based Matching System is aided by robust facial feature localization that extracts facial features in different illumination and environmental conditions and generates an optimal face template.

Fast enrolling speed – 500 enrollment/second per GPU
Ultra-fast matching speed – 5,00,000 matches per core per second
Matching accuracy
- ~0.01% at FAR 0.001 for ICAO standardized images
- ~0.27% at FAR 0.001 for any recognized frontal face databases (Non ICAO standard)
Field adjustable image resolution support
Support for diverse image & video formats
Low storage size - only 8 kB for a single image template

Iris Matching

Our Iris Matching algorithm recently won the first place in Mobile Iris Challenge Evaluation II (MICHE II). It supports live enrollment from a digital iris scanner. The algorithm is highly accurate in identifying low quality images, and gives live feedback on the quality of the image. Key characteristics include:

Robust iris detection
Automatic interlacing detection and correction
Correct iris segmentation
Standard Compatibility - ISO 19794-6(2005, 2010)

Voice Recognition

Decatur’s Voice recognition technology accurately identifies a speaker irrespective of the words spoken. Our proprietary algorithm implements voice enrollment and voice-print matching within different voice databases. The solution can be effectively integrated with other biometric technologies such as facial or fingerprint recognition systems.

Supports 1:1 matching and 1:N identification
Employs proprietary noise removal and speech feature extraction techniques
Executes artificially intelligent speech matching
Demonstrates fast matching speed with high accuracy rate
Requires minimal setup cost; easy to integrate with existing biometric systems
Supports both live and offline enrollment
Supports batch enrollment

Latent Fingerprint/Palm print

Latent fingerprint/palm print verification is useful in forensic and Criminal Investigation systems. Our latent print image enhancement tool simplifies the verification and editing process including finger/palm print rotation, image reversal, contrast/brightness adjustments and comprehensive search and review.