ISO/IEC 29794-6:2015

INTERNATIONAL ISO/IEC
STANDARD 29794-6
First edition
2015-07-01
Information technology — Biometric
sample quality —
Part 6:
Iris image data
Technologies de l’information — Qualité d’échantillon biométrique —
Partie 6: Image d’iris
Reference number
©
ISO/IEC 2015
© ISO/IEC 2015, Published in Switzerland
All rights reserved. Unless otherwise specified, no part of this publication may be reproduced or utilized otherwise in any form
or by any means, electronic or mechanical, including photocopying, or posting on the internet or an intranet, without prior
written permission. Permission can be requested from either ISO at the address below or ISO’s member body in the country of
the requester.
ISO copyright office
Ch. de Blandonnet 8 • CP 401
CH-1214 Vernier, Geneva, Switzerland
Tel. +41 22 749 01 11
Fax +41 22 749 09 47
copyright@iso.org
www.iso.org
ii © ISO/IEC 2015 – All rights reserved

Contents Page
Foreword .v
Introduction .vi
1 Scope . 1
2 Conformance . 1
3 Normative references . 2
4 Terms and definitions . 2
5 Acronyms and abbreviated terms. 3
6 Iris image quality metrics . 3
6.1 General . 3
6.2 Required iris image quality metrics computed from a single image . 4
6.2.1 Usable iris area . 4
6.2.2 Iris-sclera contrast . 5
6.2.3 Iris-pupil contrast . 6
6.2.4 Pupil boundary circularity . 7
6.2.5 Grey scale utilisation . 8
6.2.6 Iris radius . 8
6.2.7 Pupil dilation . 9
6.2.8 Iris pupil concentricity . 9
6.2.9 Margin adequacy.10
6.2.10 Sharpness .12
6.3 Recommended iris image quality metrics computed from a single image .13
6.3.1 Frontal gaze–elevation .13
6.3.2 Frontal gaze-azimuth .13
6.3.3 Motion blur .15
6.4 Iris image quality metrics computed from two images .15
6.4.1 Common usable iris area .15
6.4.2 Dilation constancy .15
6.4.3 Illumination similarity .16
6.5 Unified (overall) quality score .16
6.5.1 General.16
6.5.2 Computational method.16
7 Iris acquisition quality .17
7.1 General .17
7.2 Dedicated illumination .17
7.2.1 Description . . .17
7.2.2 Units of measure .17
7.2.3 Computational method.18
7.2.4 Value range/threshold .18
7.3 Modulation transfer function .18
7.3.1 Description . . .18
7.3.2 Units of measure .18
7.3.3 Computational method.18
7.3.4 Value range/threshold .18
7.4 Spatial sampling rate .18
7.4.1 Description . . .18
7.4.2 Units of measure .19
7.4.3 Computational method.19
7.4.4 Value range/threshold .19
7.5 Optical distortion .19
7.6 Pixel aspect ratio .19
7.6.1 Description . . .19
7.6.2 Units of measure .19
© ISO/IEC 2015 – All rights reserved iii

7.6.3 Computational method.19
7.6.4 Value range/threshold .19
7.7 Sensor signal-to-noise ratio .19
7.7.1 Description . . .19
7.7.2 Units of measure .19
7.7.3 Computational method.20
7.7.4 Value range/threshold .20
8 Iris image quality data record .20
8.1 Binary encoding .20
8.2 XML encoding .22
Annex A (normative) Conformance test assertions .24
Annex B (informative) Iris image quality .25
Bibliography .29
iv © ISO/IEC 2015 – All rights reserved

Foreword
ISO (the International Organization for Standardization) and IEC (the International Electrotechnical
Commission) form the specialized system for worldwide standardization. National bodies that are
members of ISO or IEC participate in the development of International Standards through technical
committees established by the respective organization to deal with particular fields of technical
activity. ISO and IEC technical committees collaborate in fields of mutual interest. Other international
organizations, governmental and non-governmental, in liaison with ISO and IEC, also take part in the
work. In the field of information technology, ISO and IEC have established a joint technical committee,
ISO/IEC JTC 1.
The procedures used to develop this document and those intended for its further maintenance are
described in the ISO/IEC Directives, Part 1. In particular the different approval criteria needed for
the different types of document should be noted. This document was drafted in accordance with the
editorial rules of the ISO/IEC Directives, Part 2 (see www.iso.org/directives).
Attention is drawn to the possibility that some of the elements of this document may be the subject
of patent rights. ISO and IEC shall not be held responsible for identifying any or all such patent rights.
Details of any patent rights identified during the development of the document will be in the Introduction
and/or on the ISO list of patent declarations received (see www.iso.org/patents).
Any trade name used in this document is information given for the convenience of users and does not
constitute an endorsement.
For an explanation on the meaning of ISO specific terms and expressions related to conformity
assessment, as well as information about ISO’s adherence to the WTO principles in the Technical Barriers
to Trade (TBT) see the following URL: Foreword - Supplementary information
The committee responsible for this document is ISO/IEC JTC 1, Information technology, SC 37, Biometrics.
ISO/IEC 29794 consists of the following parts, under the general title Information technology — Biometric
sample quality:
— Part 1: Framework
— Part 4: Finger image data
— Part 5: Face image data (Technical Report)
— Part 6: Iris image data
ISO/IEC 29794 will be prepared to accommodate new, additional parts that address other modalities
specified by ISO/IEC 19794, with part numbers and titles aligning appropriately.
© ISO/IEC 2015 – All rights reserved v

Introduction
The assessment of biometric sample quality through the calculation of quality metrics can be used to
predict the resulting identification accuracy in the framework of a given biometric system. With proper
use, quality metrics can enhance the functionality of a biometric system. For example they can provide
feedback regarding the integrity of collected biometric data during the enrolment or identification
process.
The purpose of this part of ISO/IEC 29794 is to define terms and quantitative methodologies relevant
to characterizing the quality of iris images and to assess their potential for high confidence biometric
match decisions.
ISO/IEC 19784-1 and ISO/IEC 19785-1 standards allocate a quality field and specify a quality score
range applicable to iris images with a qualitative foundation. ISO/IEC 19794-6 includes an informative
annex covering the subject of iris image capture and provides image quality guidelines. However, these
International Standards do not contain specific content to guide the quantitative formation of iris image
quality metrics or the interpretations of such metrics. This part of ISO/IEC 29794 establishes required
ranges of covariate values where definitive empirical data exists to justify such ranges. In other cases,
ranges of covariate values are specified as non-normative recommendations.
This part of ISO/IEC 29794 is structured as follows. The first five Clauses state Scope, Conformance,
Normative references, Terms and definitions, and Acronyms. Clause 6 specifies a set of quality metrics
for assessing the quality of iris images. Some of the metrics are declared as normative, as their impacts
on recognition rates have been quantified, while others are only informative, allowing their use as they
may provide valuable information for further stages in the biometric system. Some of the metrics in
Clause 6 are applicable to the analysis of single images, while others are applicable to assessing the
utility of a given pair of images for mutual comparison.
Clause 7 is dedicated to provide guidance to acquisition device manufacturers by defining quality
parameters that shall be considered for generating conformant iris images.
Clause 8 establishes encoding of the iris image quality data record.
vi © ISO/IEC 2015 – All rights reserved

INTERNATIONAL STANDARD ISO/IEC 29794-6:2015(E)
Information technology — Biometric sample quality —
Part 6:
Iris image data
1 Scope
This part of ISO/IEC 29794 establishes
— methods used to quantify the quality of iris images,
— normative requirements on software and hardware producing iris images,
— normative requirements on software and hardware measuring the utility of iris images,
— terms and definitions for quantifying iris image quality, and
— standardized encoding of iris image quality.
Outside the scope is
— performance evaluation of specific iris quality assessment algorithms.
2 Conformance
An iris image shall be of sufficient utility if the measurements required by 6.2.X.3 satisfy the valid
range/thresholds specified in 6.2.X.4.
A pair of images of an iris shall be of sufficient utility if the pair conforms to the requirements of 6.4.
Specifically, they shall satisfy valid range/thresholds specified in 6.4.X.4 using computation method
specified in 6.4.X.3.
An iris image quality record shall conform to this part of ISO/IEC 29794 if its structure and data values
conform to the formatting requirements of Clause 8 (Iris image quality data record) and its quality
values are computed using the methods specified in 6.2.X.3. Conformance to the normative requirements
of Clause 8 fulfils Level 1 and Level 2 conformance as specified in ISO/IEC 19794-1:2011, Annex A.
Conformance to the normative requirements of Clause 6.2.X.3 is Level 3 conformance as specified in
ISO/IEC 19794-1:2011, Annex A.
An iris acquisition device shall conform to this part of ISO/IEC 29794 if it conforms to the normative
requirements of Clause 7.
Computation of the utility of an iris image shall conform to the requirements of 6.2, specifically the
computation methods described in 6.2.X.3. Computation of the utility of the pair of images shall be
assessed per normative requirements of 6.4, specifically the computation methods described in 6.4.X.3.
If an implementation of the metrics in this part of ISO/IEC 29794 reports an unacceptable (low) quality
value for one or more quality metrics, another image of the subject should be re-captured. This should
be repeated until
— a fully conformant image has been acquired, or
— it is determined that repeated acquisitions will not yield a sufficient quality (e.g., correct enrolment)
within the application time constraint. In this case, one unacceptable image is chosen and retained
as the best possible candidate.
© ISO/IEC 2015 – All rights reserved 1

3 Normative references
The following documents, in whole or in part, are normatively referenced in this document and are
indispensable for its application. For dated references, only the edition cited applies. For undated
references, the latest edition of the referenced document (including any amendments) applies.
ISO/IEC 19794-1:2011, Information technology — Biometric data interchange formats — Part 1: Framework
ISO/IEC 19794-6:2011, Information technology — Biometric data interchange formats — Part 6: Iris image
data
ISO/IEC 29794-1, Information technology — Biometric sample quality–Part 1: Framework
4 Terms and definitions
For the purpose of this document, the terms in ISO/IEC 19794-6:2011, ISO/IEC 29794-1, and the following
apply.
4.1
covariate
variable or parameter that either directly, or when interacting with other covariates, affects iris
recognition accuracy
Note 1 to entry: Synonyms are variable, explanatory variable, and quality parameter.
Note 2 to entry: Accuracy might be stated in terms of false negative identification rate, false positive identification
rate, false non-match rate, false match rate, failure-to-enrol rate, or failure-to-acquire rate.
4.2
defocus
image impairment due to the position of the iris along the optical axis of the camera away from the plane
or surface of best focus, generally resulting in reduced sharpness (blur) and reduced contrast
4.3
depth of field
a distance range relative to the entrance aperture of a capture device over which the iris image has
greater than a specified quality with respect to focus
4.4
iris centre
centre of a circle approximating the boundary between the iris and the sclera
4.5
iris radius
radius of a circle approximating the boundary between the iris and the sclera
4.6
metric
quantification of a covariate using a prescribed method
4.7
modulation
waveform with maximum and minimum values, max and min, 100(max-min)/(max+min)%
4.8
modulation transfer function
ratio of the image modulation to the object modulation at specified spatial frequencies
2 © ISO/IEC 2015 – All rights reserved

4.9
normalised image
iris portion of the image that is mapped into doubly-dimensionless polar coordinates in which the radial
coordinate between the inner and outer boundaries of the iris along any angular ray from the iris centre
is normalised to lie between 0 and 1, in order to impart both size invariance for the imaged iris and also
invariance to pupil dilation
4.10
pupil centre
centre of a circle approximating the boundary between the iris and the pupil
Note 1 to entry: This definition gives a more robust estimate of pupil centre than the definition in ISO/IEC 19794-
6:2011 because it is less sensitive to occlusions on the iris pupil boundary. ISO/IEC 19794-6:2011 defines pupil
centre as the average of coordinates of all the pixels lying on the boundary of the pupil and the iris.
4.11
segmentation
process of determining, within an image containing an iris, the boundaries between areas containing
visible iris tissue and those that do not
Note 1 to entry: This process is preceded by localisation of the iris, and typically followed by cropping or masking
regions that are not iris tissue.
4.12
spatial sampling rate
number of picture elements (pixels) per unit distance in the object plane or per unit angle in the imaging
system
5 Acronyms and abbreviated terms
MTF Modulation Transfer Function
6 Iris image quality metrics
6.1 General
This Clause establishes requirements for assessing the quality of an iris image (Clause 6.2 and 6.3)
and pairs of iris images to be compared (Clause 6.4). Image quality metrics computed from a single
image (quality metrics hereafter) are useful to ensure the acquired images are suitable for biometric
comparison. Image quality metrics computed from a pair of images (mutual quality metrics hereafter)
are useful to ensure the reliability of the outcome when comparing the two images. Mutual quality
metrics indicate how the difference of image-specific covariates between two iris images affect their
expected comparison scores.
Clause 6.2 specifies the normative quality requirements for an iris image of sufficient utility. Quality
[11]
metrics in Clause 6.2 are ordered in terms of their effects on recognition error rates, such that the
one with the largest effect on recognition performance is listed first.
Clause 6.3 specifies recommended quality requirements for an iris image. These quality metrics
have been reported to affect recognition accuracy, but either their effect on recognition accuracy or
the methods for computing them have not been quantitatively verified to be reliable or interoperable.
Therefore, these metrics are not considered normative in the scope of this part of ISO/IEC 29794.
Clause 6.4 specifies normative requirements for mutual quality metrics including units of measurement,
the method of computation, and the acceptable range of mutual quality metrics of the two iris images to
be compared.
Required or recommended values or bounds in Clauses 6.2.X.4 and 6.4.X.4 are based on currently available
[11][12]
empirical studies. If an implementation of the metrics in this part of ISO/IEC 29794 reports an
© ISO/IEC 2015 – All rights reserved 3

unacceptable (low) quality value for one or more quality metrics, another image of the subject should
be re-captured. This should be repeated until either a fully conformant image has been acquired, or it is
determined that repeated acquisitions will not yield a sufficient quality (e.g., correct enrolment) within
the application time constraint. In this case, one unacceptable image is chosen and retained as the best
possible candidate. A NOTE at the end of each 6.2.X.4 and 6.3.X.4 sub-clauses instruct an enrolment
official on how to remedy the problem.
Informative Annex B provides information on iris image covariates that are influential on image quality
and hence recognition accuracy. It distinguishes between iris covariates based on the fixed design
parameters of the acquisition device or the operation of the device (Clause B.2 Iris acquisition covariates)
and subject covariates (Clause B.3).
6.2 Required iris image quality metrics computed from a single image
6.2.1 Usable iris area
6.2.1.1 Description
USABLE_IRIS_AREA is the fraction of the iris portion of the image that is not occluded by eyelids,
eyelashes, or specular reflections. USABLE_IRIS_AREA shall be computed as the non-occluded fraction
of the area between two circles approximating iris-sclera and iris-pupil boundaries, expressed as a
percentage.
Patterned contact lenses hide iris tissue and should be avoided.
NOTE 1 Figure 1 shows examples of iris images with various occlusions.
NOTE 2 Usable iris area computed for a single image is important for ensuring that images are of adequate
utility. Therefore, a subject enrolment process has to aim for maximising this covariate for the individual
concerned. Meanwhile, estimating the common usable iris area in the context of two iris images to be compared
is also valuable, since the image area used for biometric comparison consists of regions that are not occluded in
either image. See Clause 6.4.1.
6.2.1.2 Units of measure
USABLE_IRIS_AREA is dimensionless.
6.2.1.3 Computational method
USABLE_IRIS_AREA shall be measured following iris segmentation and after locating all occluded pixels
in the iris portion of the image using the procedure below:
1. Approximate iris-sclera and iris-pupil boundaries as two circles.
2. Denote N as the count of the pixels between the two circles.
iris
3. Denote N as the count of the pixels between the two circles that are occluded by eyelids,
occluded
eyelashes, or specular reflections.
4. Compute USABLE_IRIS_AREA as follows:
 
N
occluded
1− ×100
 
N
 iris 
NOTE Regions of the iris occluded by eyelashes may be excluded by applying a threshold to the histogram of
[2]
the pixels in the segmented iris portion of the image between the detected eyelids.
4 © ISO/IEC 2015 – All rights reserved

(a) Eye lashes (b) Eye lid (c) Eyeglass frame (d) Specularities
Figure 1 — Example images with different occlusions
6.2.1.4 Value range/threshold
[11]
USABLE_IRIS_AREA shall be 70 or larger.
The presence of an artifice such as patterned contact lenses should be detected and if detected it shall
be recorded in the quality record (see Table 2) and shall be included as an occlusion in computation of
USABLE_IRIS_AREA (Step 3 in 6.2.1.3).
NOTE If an image has unacceptable USABLE_IRIS_AREA, further images might be collected after the subject
has been asked to open the eyes more widely, to push away long eye lashes, and to look directly into the camera.
6.2.2 Iris-sclera contrast
6.2.2.1 Description
IRIS_SCLERA_CONTRAST represents the image characteristics at the boundary between the iris region
and the sclera. Sufficient contrast is needed in many implementations of iris segmentation algorithms.
Low or insufficient contrast may result in a failure to process an iris image during feature extraction.
NOTE 1 The intrinsic iris-sclera contrast varies among human irises. Iris-sclera contrast of an iris image is
affected by both the intrinsic contrast and extrinsic conditions such as illumination wavelength and other capture
device characteristics.
NOTE 2 This metric is different from GREY_SCALE_UTILISATION.
6.2.2.2 Units of measure
IRIS_SCLERA_CONTRAST is dimensionless, expressed as a percentage.
6.2.2.3 Computational method
IRIS_SCLERA_CONTRAST shall be computed as follows:
1. Approximate iris-sclera and iris-pupil boundaries as two circles.
2. Normalise so that iris-sclera boundary is at a radius of 1,0.
3. Select all pixels in an annulus whose outer radius is 0,9 and whose inner radius extends to the
midpoint between iris-pupil and iris-sclera boundaries, which are not occluded by eyelids, eyelashes,
specular reflections, or boundaries of hard contact lenses. Let these be termed iris pixels.
4. Set iris_value as the median of iris pixels.
5. Select all pixels that are not occluded by eyelids, eyelashes, or specular reflections in an annulus
with inner radius of 1,1 and outer radius of 1,2. Let these be termed sclera pixels.
© ISO/IEC 2015 – All rights reserved 5

6. Set sclera_value as the median of sclera pixels.
7. IRIS_SCLERA_CONTRAST
0 pupilv__alue≥≥iris valueORpupil__valuescleravalue




=

sclera_vaalue−iris_value

×100 ottherwise

sclera__valuei+−risvalue 2×pupilv_ alue


NOTE This computation can proceed even if the approximating iris-sclera and iris-pupil circles are not
concentric.
Pupil_value is defined in 6.2.3.3.
6.2.2.4 Value range/threshold
IRIS_SCLERA_CONTRAST shall be larger than 5.
NOTE If an image has unacceptable IRIS_SCLERA_CONTRAST, another image might be captured, perhaps
after moving away from extraneous light. If the problem persists an alternative camera might be used. Generally,
increased illumination or camera gain may help improve IRIS_SCLERA_CONTRAST.
6.2.3 Iris-pupil contrast
6.2.3.1 Description
IRIS_PUPIL_CONTRAST represents the image characteristics at the boundary between the iris region
and the pupil. Sufficient iris-pupil contrast is needed in many implementations of iris segmentation
algorithms. Low or insufficient contrast may result in a failure to process an iris image during feature
extraction.
NOTE 1 The intrinsic iris-pupil contrast varies among human irises. Iris-pupil contrast of an iris image is
affected by both the intrinsic contrast and extrinsic conditions such as illumination wavelength and other capture
device characteristics.
NOTE 2 Intrinsic contrast may be different between visible light illuminated iris images and near infrared
illuminated iris images.
NOTE 3 This metric is different from GREY_SCALE_UTILISATION.
6.2.3.2 Units of measure
IRIS_PUPIL_CONTRAST is dimensionless.
6.2.3.3 Computational method
IRIS_PUPIL_CONTRAST shall be computed as follows:
1. Approximate iris-pupil and iris-sclera boundaries as two circles.
2. Normalize so that iris-pupil boundary is at a radius of 1,0.
3. Select all pixels inside a circle of radius 0,8 that are not occluded by eyelids, eyelashes, or specular
reflections. Let this be denoted as pupil pixels.
4. Set pupil_value as the median of pupil pixels.
6 © ISO/IEC 2015 – All rights reserved

5. Select all pixels in an annulus whose inner radius is 1,1 and whose outer radius extends to the
midpoint between iris-pupil and iris-sclera boundaries, which are not occluded by eyelids, eyelashes,
specular reflections, or boundaries of hard contact lenses. Let these be termed iris pixels.
6. Set iris_value as the median of iris pixels.
iris__valuep− upil value
7. Compute weberr_ atio=
20+pupilv_ alue
weberr_ atio
8. IRIS_PUPIL_CONTRAST= ×100
07._5+weberratio
NOTE 1 Eyelashes can occlude pupil as shown in Figure1a.
NOTE 2 If the pupil is black, then the definition of (Michelson) contrast of the IRIS_PUPIL_CONTRAST will
always be 100%. Therefore, normalised Weber contrast is more informative.
6.2.3.4 Value range/threshold
IRIS_PUPIL_CONTRAST shall be 30 or more.
NOTE If an image has unacceptable IRIS_PUPIL_CONTRAST, another image might be captured, perhaps
after moving away from extraneous light. If the problem persists an alternative camera might be used. Generally,
increased illumination or camera gain may help improve IRIS_PUPIL_CONTRAST.
6.2.4 Pupil boundary circularity
6.2.4.1 Description
PUPIL_BOUNDARY_CIRCULARITY represents the circularity of the iris-pupil boundary.
NOTE 1 Deviation from circularity in the iris-pupil boundary can affect segmentation accuracy. The effect
of this metric on performance depends on the sensitivity of the segmentation algorithm to the deviation from
circularity in iris-pupil boundaries.
NOTE 2 The non-circularity could be due either to natural anatomical variation or to non-frontal gaze or both.
NOTE 3 Certain medical conditions or treatments can induce highly non-circular pupils.
6.2.4.2 Units of measure
PUPIL_BOUNDARY_CIRCULARITY is dimensionless.
6.2.4.3 Computational method
PUPIL_BOUNDARY_CIRCULARITY shall be measured by the total modulus (sum of the squared
coefficients) of the real and imaginary parts of a Fourier series expansion of the pupil boundary, in
[2]
radius as a function of angle around the centre. Specifically:
1. Compute a Fourier expansion of N regularly spaced angular samples of radial gradient edge data
{r } for θ = 0 to N – 1 spanning [0,2π].
θ
A set of M discrete Fourier coefficients {C }, k = 0, …, M-1, where M is much smaller than N, is derived
k
N−1
−2/πθik N
from data sequence {r } as Cr= e
θ
k ∑ θ
θ=0
M−1
2. PUPIL_BOUNDARY_CIRCULARITY=max(,0100− C )
k
∑
N
k=1
© ISO/IEC 2015 – All rights reserved 7

The above calculation shall include only the non-occluded pixels on the boundary of iris and pupil. M is
the number of activated Fourier coefficients and specifies the degrees of freedom in the shape model. M
[2]
should be set to 17 to capture the true pupil boundary with appropriate fidelity.
NOTE In the case of a truly circular boundary, all frequency coefficients higher than DC (the zeroth-term
C ) in this Fourier series would be 0, and the C /N is exactly the radius of that perfect circle. To the extent that
0 0
frequency components higher than the DC term have non-zero coefficients, the boundary is non-circular. The
total modulus measures this non-circularity.
6.2.4.4 Value range/threshold
PUPIL_BOUNDARY_CIRCULARITY will be 100 for a circle and [0,100) for anything else.
NOTE The non-circular pupil may be innate to the eye being imaged and may therefore not be remediable.
6.2.5 Grey scale utilisation
6.2.5.1 Description
GREY_SCALE_UTILISATION examines pixel values of an iris image for evidence of a spread of intensity
values in iris portion of the image. A useful iris image should have a dynamic range of 256 grey levels,
allocating at least 8 bits with a minimum of 6 bits of useful information. An “underexposed” image
would have too few high intensity pixels, and conversely for “overexposed”. An image with a high score
indicates a properly exposed image, with a wide, well distributed spread of intensity values.
6.2.5.2 Units of measure
GREY_SCALE_UTILISATION shall be measured in bits.
6.2.5.3 Computational method
For each grey level i present in the image, compute its probability p of occurring. Thus p is the total
i i
count of pixels at grey level i, divided by the total number of pixels in the image. The entropy H of the
pixel histogram, in bits, is:
Hp=− log p
∑ i 2 i
i
NOTE When xx→→00,log x so pixel values that never occur in an image can be ignored in the
calculation of entropy.
6.2.5.4 Value range/threshold
Entropy of the pixel histogram shall be 6 bits or higher.
NOTE If an image has unacceptable GREY_SCALE_UTILIZATION, another image might be captured, perhaps
after moving away from extraneous light. If the problem persists an alternative camera might be used. Generally,
increased illumination or camera gain may help improve GREY_SCALE_UTILIZATION.
6.2.6 Iris radius
6.2.6.1 Description
IRIS_RADIUS is the radius of a circle approximating the iris-sclera boundary.
[10]
NOTE The average human iris radius is 5,9 millimetres with a reported range of 5,1 to 6,5 millimetres.
8 © ISO/IEC 2015 – All rights reserved

6.2.6.2 Units of measure
IRIS_RADIUS shall be measured in pixels.
6.2.6.3 Computational method
IRIS_RADIUS shall be measured as the radius of a circle approximating the boundary between the iris
and the sclera.
NOTE This metric should be computed after segmentation.
6.2.6.4 Value range/threshold
IRIS_RADIUS shall be at least 80 pixels for the smallest reported human iris of 5,1 millimetre radius.
6.2.7 Pupil dilation
6.2.7.1 Description
PUPIL_IRIS_RATIO represents the degree to which the pupil is dilated or constricted.
6.2.7.2 Units of measure
PUPIL_IRIS_RATIO is dimensionless.
6.2.7.3 Computational method
PUPIL_RADIUS
PUPIL_IRIS_RATIO=×100
IRIS_RADIUS
PUPIL_RADIUS shall be measured as the radius of the circle approximating pupil shape. The DC (zeroth-
term C ) in the Fourier series expansion of pupil boundary (Clause 6.2.4.3) approximates radius of the
pupil.
IRIS_RADIUS shall be measured as specified in Clause 6.2.6.3.
NOTE This metric should be computed after segmentation.
6.2.7.4 Value range/threshold
PUPIL_IRIS_RATIO shall be between 20 and 70.
NOTE 1 This is the only quality metric for which the higher-the-better rule for quality does not apply.
NOTE 2 If pupil dilation is not within the desired range, another image might be captured after adjusting the
ambient light levels to decrease or increase dilation.
6.2.8 Iris pupil concentricity
6.2.8.1 Description
IRIS_PUPIL_CONCENTRICITY represents the degree to which the pupil centre and the iris centre are in
the same location.
NOTE Pupil and iris are naturally never exactly concentric. Pronounced deviation from concentricity can
cause segmentation error. Conversely, pronounced measured non-concentricity can indicate a segmentation
error.
© ISO/IEC 2015 – All rights reserved 9

6.2.8.2 Units of measure
IRIS_PUPIL_CONCENTRICITY is dimensionless.
6.2.8.3 Computational method
The value for iris and pupil concentricity shall be computed using the Euclidean distance between the
iris and pupil centres divided by the iris radius, as given by:
XX− +−YY
() ()
pupilirispupil iris
100×−max{10,}
IRIS_RADIUS
(X , Y ) and (X , Y ) are the coordinates of iris centre and pupil centre respectively.
iris iris pupil pupil
NOTE This metric should be computed after segmentation.
6.2.8.4 Value range/threshold
IRIS_PUPIL_CONCENTRICITY shall be 90 or more.
NOTE Non-concentric iris and pupil may be innate to the eye being imaged and may therefore not be
remediable. Another image might be collected after the subject has been asked to look directly into the camera
and to open their eyes widely.
6.2.9 Margin adequacy
6.2.9.1 Description
MARGIN_ADEQUACY quantifies the degree to which the iris portion of the image is centred relative
to the edges of the entire image. The maximum value for this metric shall occur when the margin
requirements of ISO/IEC 19794-6:2011 are satisfied.
6.2.9.2 Units of measure
MARGIN_ADEQUACY is dimensionless.
6.2.9.3 Computational method
There are four individual margin values: LEFT_MARGIN, RIGHT_MARGIN, UP_MARGIN and DOWN_
MARGIN.
The individual margin values shall be computed as:
X − IRIS_RADIUS
iris
LM=
IRIS_RADIUS
IMAGE_WIDTH - (IX + RISS_RADIUS)
iris
RM=
IRIS_RADIUS
IMAGE_HEIGHT - (YI+ RIS_RADIUS)
iris
DM=
IIRIS_RADIUS
Y − IRIS_RADIUS
iris
UM=
IRIS_RADIUS
10 © ISO/IEC 2015 – All rights reserved

 
 LM
LEFT_MARGIN = max 01,min ,
  
06,
 
 
 
 RM
RIGHT_MARGIN = maxx 01,min ,
  
06,
 
 
 UM 
 
UP_MARGIN = max 01,min ,
  
02,
 
 
 
 DM
DOWN_MARGIN = max 01,min ,
  
02,
 
 
(X , Y ) are iris centre in the horizontal and vertical directions where (0,0) is the top left corner of the
iris iris
image (See Figure 2). IMAGE_WIDTH, IMAGE_HEIGHT, and IRIS_RADIUS are measured in pixels.
The margin adequacy value shall be computed as:
MARGIN_ADEQUACY=100×min{LEFT_MARGIN, RIGHT_MARGIN, UP_MARGIIN, DOWN_MARGIN}
NOTE This metric could be computed after coarse segmentation.
Up Margin gin >>> 0.2 0.2 RR Up Marginn
< 0.2 R
(0,0))
R
Left Margin
<0.6 R
Left Margin Down Margin Right Margin
> 0.6 R > 0.2 R > 0.6 R
(a) An iris image with adequate margins. (b) An iris image where LEFT_MARGIN and UP_
Note that (0,0) is at the upper left corner of the MARGIN are inadequate.
image.
Figure 2 — Example of iris images with adequate and inadequate margins.
6.2.9.4 Value range/threshold
MARGIN_ADEQUACY shall be greater than 80.
NOTE 1 A value of 100 indicates that all four margin values conform to the requirements established in
ISO/IEC 19794-6:2011 namely that the margin between the iris boundary and its closest edge of the image shall
be at least 0,6 × IRIS_RADIUS in the horizontal direction and 0,2 × IRIS_RADIUS in the vertical direction.
NOTE 2 If an image has unacceptable MARGIN_ADEQUACY another image might be collected after the subject
has been asked to look directly into the camera
© ISO/IEC 2015 – All rights reserved 11

6.2.10 Sharpness
6.2.10.1 Description
[3][5][6][7]
SHARPNESS measures the degree of focus present in the image. Sharpness is measured as a
function of the power spectrum after filtering with a Laplacian of Gaussian operator.
6.2.10.2 Units of measure
SHARPNESS is dimensionless.
6.2.10.3 Computational method
Calculation of the sharpness of an image is determined by the power resulting from filtering the image
with a Laplacian of Gaussian kernel. The standard deviation of the Gaussian is 1,4.
1. The convolution kernel (F) is defined thus:
F =
01 1 222 11 0
12 4 555 42 1
145 303 54 1
25 31−−2241− 23 52
25 02−−4402− 40 5 22
25 31−−2241− 23 52
145 303 54 1
12 4 555 42 1
01 1 222 11 0
2. If I(x,y) is the image, then the weighted sum of I(x,y) as per F is computed for every fourth row and
column location in
...