Scribd
Upload
58 views

NIST Fingerprint Image Quality

NIST Fingerprint Image Quality defines fingerprint image quality as a prediction of a fingerprint matching system's performance. NIST developed the NFIQ algorithm to assess quality based on 11 image features and classify prints into 5 quality levels from excellent to poor. Evaluation showed NIST FIQ levels predict matching performance and the levels are statistically separable, making it useful for applications like sample filtering or processing. NFIQ assessment is fast and has been tested across multiple fingerprint matchers and datasets.

Uploaded by

aravind1890
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
58 views

NIST Fingerprint Image Quality

NIST Fingerprint Image Quality defines fingerprint image quality as a prediction of a fingerprint matching system's performance. NIST developed the NFIQ algorithm to assess quality based on 11 image features and classify prints into 5 quality levels from excellent to poor. Evaluation showed NIST FIQ levels predict matching performance and the levels are statistically separable, making it useful for applications like sample filtering or processing. NFIQ assessment is fast and has been tested across multiple fingerprint matchers and datasets.

Uploaded by

aravind1890
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 13

NIST Fingerprint Image Quality

Elham Tabassi
Biometric Consortium Conference
September 20, 2005
background
o quality is important

o previous research
• mostly defined as a measure of the extractability of the
features used for recognition such as minutiae.
• local orientation information (Bolle et al, Shen et al, Hong et
al., …)
• global features (Hong et al, Lim & Yao, Nill & Bouzas, …)

o almost all used subjective quality assessment to


evaluate their quality algorithm
• size of fingerprint, pressure, humidity, amount of dirt, …
BCC 2005 [email protected]
quality as prediction of performance
we define fingerprint image quality as a prediction of a matcher
performance, e.g. a sample’s quality score reflects the predictive
positive or negative contribution of an individual sample to the overall
performance of a fingerprint matching system.

excellent quality
TAR samples result in
high performance

poor quality
samples result in
low performance

BCC 2005
FAR [email protected]
use of quality to improve performance
o recapture samples of insufficient quality
• pruning the poorest quality samples (1.65% of dataset) reduced EER
from .0047 t0 .0024 (sdkI - dos - ri)

o process samples differently based on their qualities

o collect relevant statistics


• compare capture devices and/or environments
• correlation among fingers
p(nfiq(ri)=5) = 0.011 p(nfiq(li)=5) = 0.016
p(nfiq(li)=5 | nfiq(ri)=5) = 0.22

o cause higher quality sample dominate fusion

BCC 2005 [email protected]


NIST Fingerprint Image Quality

quality
NFIQ number =1
=5

NFIQ’s 5 levels of quality are intended to be predictive of the


relative performance of a minutia based fingerprint matching
system.
• NFIQ=1 indicates high quality samples, so lower FMR and/or
FNMR is expected.
• NFIQ=5 indicates poor quality samples, so higher FMR and/or
FNMR is expected.

BCC 2005 [email protected]


performance target
degree of separation between a sample’s genuine and imposter distributions

s d 2 9 - v tb m a tc h a n d n o n m a tc h s c o r e s h is to g r a m

nomatch
0 .1 5 m a tc h s c o r e s
match
n o n m a tc h s c o r e s

poor
0 .1 0
excellent

0 .0 5

0 .0 0
4 36 68 100 132 164 196 228

quality of a biometric sample xi ≜ prediction of the bin its normalized


match score falls
BCC 2005 [email protected]
pair-wise quality

Q1 pairwise quality
fingerprint
similarity
H(Qalgorithm
matching
1,Q2) Qscore
12
Q2

when the enrollment sample


is of good quality and better
than that of the use phase
(search) sample, the search
sample’s quality is sufficient
to predict performance.

BCC 2005 [email protected]


NIST Fingerprint Image Quality

NFIQ
feature
extraction
neural
network
quality
number

o feature extraction: computes appropriate signal or image fidelity


characteristics and results in an 11-dimensional feature vector.

o neural network:classifies feature vectors into five classes of quality


based on various quantiles of the normalized match score
distribution.

o quality number: an integer value between 1(highest) and 5


(poorest).

BCC 2005 [email protected]


NFIQ effectiveness
o evaluation criterion is rank ROC as a function of image quality
o used various fingerprint matching algorithms and various
datasets to evaluate NFIQ
• 15 different COTS fingerprint matching algorithms
• 22 different datasets of different fingers captured by various devices
and at different operational settings
• each test dataset has 2 fingerprint images of 6000 person
o compared (TAR,FAR) of levels of quality at a fixed threshold
• as quality degrades, true accept rate decreases for all the matchers,
FAR increase for some.
o levels 2,3,4, and 5 are statistically separable.
o It takes about one third of a second to compute quality of a flat
fingerprint image.

BCC 2005 [email protected]


6000 subjects - Right index
Threshold @ (far,tar)=(0.012,0.99)
1 2 3 4 5
quality excellent veryGood good fair poor
FAR 0.0037 0.0083 0.0131 0.0216 0.0477

TAR 0.997 0.994 0.993 0.9496 0.926


separable levels of quality
For each quality level,
we calculated 95%
confidence intervals of
TARs @ FAR=0.1% for
six matchers and sixteen
datasets.
nfiq levels 2,3,4, and 5
are statistically separate.

BCC 2005 [email protected]


conclusion

o a novel definition of fingerprint image quality


o NFIQ works as a rank statistic for performance for all 330
combinations of COTS fingerprint matchers and
operational datasets tested
o NFIQ levels 2,3,4, and 5 are statistically independent
o NFIQ can be used for real-time quality assessment
o NFIQ is publicly available but subject to US export control
laws (fingerprint.nist.gov)

BCC 2005 [email protected]


thanks
[email protected]
301 975 5292

You might also like