Fingerprint recognition is a biometrics-based pattern recognition system widely used to identify or authenticate a person by determining the authenticity of the fingerprint, an impression or mark made on a surface by one finger.
A fingerprint recognition system (FRS) works because of the uniqueness of the fingerprints since no two people have identical fingerprint patterns. They are unique to everyone, even to identical twins, and do not change over time.
Because a fingerprint is an individual characteristic, remains unchanged during an individual’s lifetime, and has general ridge patterns that can be classified systematically based on pattern types, size, and position of the finger, it is a widely used biometric characteristic.
A fingerprint-based biometric system is called either a verification system or an identification system, depending on the context. A verification system authenticates one’s identity by comparing the fingerprints to a pre-stored biometric template(s). It conducts a one-to-one comparison to determine whether the identity claimed by the individual is true. Meanwhile, an identification system recognizes an individual by searching the entire template database for a match. It conducts one-to-many comparisons to establish the identity of the individual.
With the widespread use of biometric scanners on smartphones and a growing number of security breaches, fingerprint recognition is increasingly replacing traditional authentication methods such as passwords and PINs. Fingerprint recognition offers an alternative solution to the task of personal authentication or identification since unlike passwords or biometric traits cannot be forged, lost, or forgotten.
A typical biometric system consists of four modules: a sensor module, a feature extraction module, a template database, and a matching module. The sensor module acquires the biometric image, from which a set of global or local features are then extracted by the feature extraction module. These features are stored in the template database as template data. Finally, a matching module compares the query and template data to arrive at a match or non-match verdict.
Now, a typical biometric system carries out authentication in two stages: enrollment and verification. In enrollment, a user presents the fingerprint image to be acquired by the sensor module. Certain features are extracted, and further adapted or transformed to generate template data for comparison in the next verification stage. In verification, the fingerprint image of a query is collected by the sensor module and then compared with the existing template data.
Similarly, a fingerprint verification system has four main modules such as;
- fingerprint sensing that acquires the fingerprint of an individual through a fingerprint scanner, producing a raw digital representation. Three families of electronic fingerprint sensors (such as solid-state or silicon sensors, optical, and ultrasound) are used for the acquisition of fingerprint images. Today, we also have a new generation of touchless live scan devices that generate a 3D representation of fingerprints.
- preprocessing that simplifies the input fingerprint for feature extraction.
- feature extraction, where the fingerprint is processed to generate discriminative properties, called feature vectors.
- matching, in which the feature vector is compared against existing templates stored in a database.
There are currently various types of fingerprint matching techniques that are responsible for the recognition of fingerprint patterns. A better matching technique provides a better result during matching. Some of the fingerprint matching techniques and identification keys are:
- Minutiae based
- Pattern based
- Feature based
- ROI (Region of Interest) based
- Correlation based
- Statistics based
- ONNC (Optical Neural Network Computer)
The main parameters characterizing a digital fingerprint are as follows.
Resolution: It indicates the number of dots or pixels per inch (dpi). Though 500 dpi is the minimum resolution for FBI-compliant scanners, 250 to 300 dpi is the minimum resolution needed for extraction algorithms to locate the minutiae in fingerprint patterns.
Area: It is the size of the rectangular area sensed by a fingerprint scanner. The bigger the area, the more ridges, and valleys are captured, and the more the fingerprint becomes distinctive.
Dynamic range (or depth): It denotes the number of bits used to encode each pixel’s intensity value. The FBI standard for depth of the pixel bit is 8 bits, yielding 256 gray levels.
Geometric accuracy: It is specified by the maximum geometric distortion introduced by the acquisition device, and expressed as a percentage concerning x and y directions.
Image quality: It is not easy to precisely define the quality of a fingerprint image. Image quality is key to identify the intrinsic finger status. If the fingers are too moist or too dry, or when they are incorrectly presented, most of the scanners produce poor quality images.
Benefits of FRS
- High uniqueness, accuracy, performance, and stability.
- Easy to use and reliable than other techniques.
- Good performance, low cost, and small size.
- Have approximately 50% popularity in the market as compared to other techniques.
- Has an irreplaceable role of latent prints in crime investigation.
- Enable government systems to use large databases of fingerprints with a criminal history.
Several factors influence the quality of a fingerprint image. They include skin conditions (e.g., dryness, wetness, dirtiness, temporary or permanent cuts and bruises), sensor conditions (e.g., dirtiness, noise, size), user cooperation, etc. Poor quality images result in spurious and missed features that degrade the overall system performance. Therefore, it is very critical for a fingerprint recognition system to estimate the quality and validity of the captured fingerprint images.
Key challenges in FRS
- Mismatching of fingerprints due to physical distortions, finger injuries and cuts.
- Vanishing or fading fingerprint due to jobs like cleaning.
- Mismatching due to displacement or rotation while scanning the finger.
- Unauthorized access due to finger plasticity or clay printing.
- Variability between impressions due to skin conditions, noise in the sensor.
- Small area fingerprint sensors, providing less information.
- Fake fingerprints i.e., clays or dummy printing.