Unique traits of fingers and hands are commonly used biometric markers. Those traits include fingerprints, finger vein recognition, hand recognition, and finger geometry. Hybrid identification systems have been developed that combine fingerprints and finger vein recognition. Artificial intelligence (AI) has recently been applied to fingerprint analysis, and it promises to increase its utility.
Fingerprint and finger vein technologies are similar. Both enroll individuals and store their fingerprint or finger vein patterns in a biometric database for identification purposes (Figure 1). Finger vein technology can be less invasive than fingerprinting. Finger vein is more accurate than fingerprint recognition and has a lower false rejection rate (FRR) and a lower false acceptance rate (FAR).
A near-infrared imager is used to implement finger vein recognition. The image captures the pattern of the blood vessels containing deoxidized hemoglobin as a series of dark lines that can be matched to a database and identify individuals.
Fingerprints are on the surface and easier to scan. As a result, a wide range of technologies are used to scan fingerprints, including optical, capacitive, ultrasonic, and thermal. They all work by scanning the ridges and valleys of a finger and comparing the resulting image to other images stored in a database.
Finger vein scanners measure subdermal traits and don’t rely on contact, while most fingerprint scanners require contact with the finger surface. That means that fingerprint scanners can require cleaning and maintenance that is not needed by finger vein scanners to ensure accuracy.
Hybrid finger biometrics
Finger vein authentication is more accurate and requires less maintenance than fingerprint authentication. However, using a finger vein approach produces a much more extensive biometric database than fingerprints.
The two technologies are suited to different application needs. Fingerprint technology is often associated with picking out an individual from a large database, while fingervein technology is used for security and limited searches in small databases.
A hybrid finger biometric security system can be more accurate and flexible than either technology used alone. The system simultaneously captures the fingerprint and vein data. The hybrid scanners are contactless and can be used with wet or dry fingers and variable bloodstreams.
Hand recognition
Hand recognition creates a handprint based on the unique physical characteristics of an individual’s hand. Like a fingerprint, the user’s identity can be verified by placing their hand on a scanner. Fingerprinting is generally more accurate, but hand recognition can be affected by the orientation of the hand and individual fingers on the scanner.
Four types of measurements are used to create a handprint: lengths, ratios, areas, and angles. They are computed based on the shape and location of specific segments in a hand (Figure 2).
What about finger geometry?
Finger geometry is not as amenable to automation as fingerprints. To implement finger geometry security, two or more fingers are placed on a template, and images of attributes like length, width, thickness, and the distance between fingers are taken. Some systems use three-dimensional imaging to improve accuracy. Taking a finger geometry reading often requires supervision.
Finger geometry has a significant drawback because it does not provide unique characteristics like fingerprints or other biometric traits. However, implementing other biometric security technologies can be less costly. As a result, finger geometry is limited to cost-sensitive environments with high volumes of individuals and less stringent security needs.
AI and fingerprints
A long-held belief has been that fingerprints of different fingers from the same person are unique and, therefore, unmatchable. An artificial intelligence AI program recently discovered a new way to analyze fingerprints that can accurately match fingerprints from different fingers from the same person. That discovery can potentially extend the accuracy and utility of traditional fingerprinting.
Summary
Fingerprints and finger vein recognition are widely used and highly accurate biometric technologies. They are suited for different application requirements and are sometimes combined into a hybrid system that provides enhanced performance. Hand recognition can provide a third option that accurately identifies individuals. Finger geometry identification is less accurate but suited for cost-sensitive applications. Finally, AI is applied to fingerprint analysis, which could significantly extend its utility.
References
A hand-based biometric system in visible light for mobile environments, Information Sciences
AI Discovers That Not Every Fingerprint Is Unique, Columbia University
Biometric modality: Finger geometry, Biometrics Institute
Feature identification and classification of hand-based biometrics through ensemble learning approach, Measurement: Sensors
Fingerprint vs. Finger-Vein: The Quest for Ideal Biometric Authentication, Bayometric
Hand-based multimodal biometric fusion: A review, Association for Computing Machinery
WTWH related links
What are the different types of fingerprint scanners?
How can a machine recognize hand gestures?
Sensor fusion levels and architectures
How can you detect hidden cameras invading your privacy?