WHAT IS A FINGERPRINT?
A fingerprint appears as a collection of ridges and furrows (valleys) (Watson, 2008).
- Dark lines = ridges
- White space = furrows (valleys)
- Loop - curves
- Whorl - spiral pattern
- Arches - narrow mountain
At the local level, there are different ways the ridge can be end, which is called minutiae (Please see Figure 2) (Jain, Flynn & Ross, 2008). Some examples include:
|
WHY FINGERPRINT BIOMETRIC?
Fingerprint recognition is popular due to:
The benefits and weakness of fingerprints include (Shoniregun & Crosier, 2008):
- Collectability - Ease of acquisition (Prabhakar, Pankanti & Jain, 2003; Jain et al., 2008)
- Universality – Every person has it and there’s plentiful sources (10 fingers) (Prabhakar et al., 2003; NSTC, n.d.)
- Distinctiveness - e.g. Identical twins have distinct fingerprints (Prabhakar et al., 2003; NSTC, n.d.; Jain et al., 2008)
- Consistency over time (permanence) - Except when accidents occur such as cuts (Prabhakar et al., 2003; Jain et al., 2008)
The benefits and weakness of fingerprints include (Shoniregun & Crosier, 2008):
COMPONENTS OF FINGERPRINT BIOMETRIC SYSTEM
DATA COLLECTION
There are several ways fingerprint can be acquired. The oldest method used by law enforcement is by using the "ink technique" where "the subject's finger was spread with black ink and pressed against a paper card; the card was then scanned by using a common paper-scanner, producing the final digital image" (Jain et al., 2008). A similar method can be seen when acquiring fingerprints from a crime scene (Jain et al., 2008).
Modern day fingerprint acquisition uses fingerprint scanners - no ink is required (Jain et al., 2008). There three main families of fingerprint scanner:
There are several issues with acquiring fingerprints. If the same process of acquiring fingerprints is repeated several times, an exact same image cannot be achieved every time due to (Shoniregun & Crosier, 2008):
There are several ways fingerprint can be acquired. The oldest method used by law enforcement is by using the "ink technique" where "the subject's finger was spread with black ink and pressed against a paper card; the card was then scanned by using a common paper-scanner, producing the final digital image" (Jain et al., 2008). A similar method can be seen when acquiring fingerprints from a crime scene (Jain et al., 2008).
Modern day fingerprint acquisition uses fingerprint scanners - no ink is required (Jain et al., 2008). There three main families of fingerprint scanner:
- Optical – This type of sensor uses light to differentiate valleys and ridges. The light is "reflected at the valleys and absorbed at the ridges" (Jain et al., 2008).
- Solid-state – This is also known as silicon sensors. Silicon sensors consists of a collection of pixels which as "tiny sensor itself" (Jain et al., 2008). It works by converting the fingerprint into electrical signals using one of four methods: capacitive, thermal, electric field and peizoelectric (Jain et al., 2008).
- Ultrasound – This type of sensor uses high frequency ultrasound to measure the distance between the finger on the plate and the air (Shoniregun & Crosier, 2008).
There are several issues with acquiring fingerprints. If the same process of acquiring fingerprints is repeated several times, an exact same image cannot be achieved every time due to (Shoniregun & Crosier, 2008):
- Sensor noise
- Dry fingers
- Temperature and humidity
- Physical characteristics e.g. cut
FEATURE EXTRACTION
A fingerprint has distinct characteristics (e.g. loops, whorls, minutiae). There are several steps in feature extraction in order to profile the subject's fingerprint. This includes:
A fingerprint has distinct characteristics (e.g. loops, whorls, minutiae). There are several steps in feature extraction in order to profile the subject's fingerprint. This includes:
- Segmentation – separates fingerprint image from the background (Jain et al., 2008)
- Singularity Detection – identify pattern typologies (Jain et al., 2008)
- Enhancement and Binarization – improve the clarity of the ridges and valleys on the area of the image that is recoverable and "mark the unrecoverable regions as too noisy for further processing"; most often by applying contextual filters (Jain et al., 2008)
- Minutiae Extraction – reduce the ridge lines to one pixel (Jain et al., 2008)
MATCHING ALGORITHMS
The next component of a biometric fingerprint system is matching a sample with the biometric template.
Matching can be a challenge due to:
There are three main matching algorithm families (Grialue Biometrics, 2014):
The next component of a biometric fingerprint system is matching a sample with the biometric template.
Matching can be a challenge due to:
- Large Intra-subject variation - difference seen in the image impression of the same finger which may be due to low quality image, different skin condition, feature extraction errors etc. (Jain et al., 2008)
- Small Inter-subject variation - similar images of a fingerprint (Jain et al., 2008)
There are three main matching algorithm families (Grialue Biometrics, 2014):
- Minutiae-based
- Input images rotated and shifted so that the minutiae (uniqueness and splits of the ridges) are aligned to the template fingerprint
- Minutiae are matching if "the spacial distance and direction difference between them are smaller that a given tolerance"
- Most widely used
- Limitations - requires high quality images to extract the minutiae and time consuming
- Correlation-based
- Input images are rotated and shifted so that the input image is aligned to the template fingerprint
- Correlation is computed based on the patterns of the ridges, shapes, breaks etc...for each pixel
- Ridge Feature Based
- A fingercode is composed of the features extracted based on the ridges
- Can be used in conjunction with minutiae-based algorithm to improve accuracy
- Matching is based on "computing the difference between the two fingercodes" (Grialue Biometrics, 2014)
Examples of Vulnerabilities in Fingerprint Biometric
Recall the attack vectors of biometric systems. Below are examples of attacks on the sensor of the fingerprint biometric system.
1) How To Copy a Fingerprint Like a Spy
1) How To Copy a Fingerprint Like a Spy
- Attack Vector - Sensor
- Technique - Spoofing
2) How to fake a fingerprint and break into a phone
- Attack Vector - Sensor
- Technique - Spoofing
3) Replay Attack
- Attack Vector - Sensor
- Technique: Replay Attack
“If the attacker can break the kernel [the core of the Android operating system], although he cannot access the fingerprint data stored in the trusted zone, he can directly read the fingerprint sensor at any time. Every time you touch the fingerprint sensor, the attacker can steal your fingerprint,” Yulong Zhang, one of the researchers, explained, to Forbes. “You can get the data and from the data you can generate the image of your fingerprint. After that you can do whatever you want” (Pagiani, 2015).
References:
- Griaule Biometrics (2014). Matching Approaches. Retrieved on June 25, 2016 from http://www.griaulebiometrics.com/en-us/book/understanding-biometrics/types/matching
- Ideartek (2014). Banner Image. Retrieved on July 1, 2016 from http://www.ideartek.com/content/?213.html
- Jain, A., Flynn, P. & Ross, A. (2008). Handbook of Biometrics. New York, NY: Springer US
- NSTC. (n.d.). Fingerprint Recognition. Retrieved on July 1, 2016 from https://www.fbi.gov/about-us/cjis/fingerprints_biometrics/biometric-center-of-excellence/files/fingerprint-recognition.pdf
- Pagiani, P. (2015). Samsung Galaxy S5 vulnerability allows hackers to steal fingerprints. Retrieved on July 8, 2016 from http://securityaffairs.co/wordpress/36326/hacking/samsung-s5-flaw-steal-fingerprints.html
- Prabhakar, S., Pankanti, S., & Jain, A. (2003). Biometric Recognition: Security and Privacy Concerns. IEEE Security & Privacy. Retrieved on July 1, 2016 from http://biometrics.cse.msu.edu/Publications/GeneralBiometrics/PrabhakarPankantiJain_BiometricSecurityPrivacy_SPM03.pdf
- Shoniregun, C. & Crosier, S. (2008). Securing Biometrics Applications. New York, NY: Springer US
- Toli, C. & Preneel, B (2015). Biometric Solutions as Privacy Enhancing Technologies. Retreived on July 8, 2016 from https://www.researchgate.net/publication/280307395_Biometric_Solutions_as_Privacy_Enhancing_Technologies
- Watson, S. (2008). How Stuff Works - How Fingerprinting Works. Retrieved on July 7, 2016 from http://science.howstuffworks.com/fingerprinting1.htm