ISO/IEC 29159-1:2010

INTERNATIONAL ISO/IEC
STANDARD 29159-1
First edition
2010-09-01
Information technology — Biometric
calibration, augmentation and fusion
data —
Part 1:
Fusion information format
Technologies de l'information — Étalonnage biométrique, données
d'augmentation et de fusion —
Partie 1: Format d'information de fusion

Reference number
©
ISO/IEC 2010
PDF disclaimer
This PDF file may contain embedded typefaces. In accordance with Adobe's licensing policy, this file may be printed or viewed but
shall not be edited unless the typefaces which are embedded are licensed to and installed on the computer performing the editing. In
downloading this file, parties accept therein the responsibility of not infringing Adobe's licensing policy. The ISO Central Secretariat
accepts no liability in this area.
Adobe is a trademark of Adobe Systems Incorporated.
Details of the software products used to create this PDF file can be found in the General Info relative to the file; the PDF-creation
parameters were optimized for printing. Every care has been taken to ensure that the file is suitable for use by ISO member bodies. In
the unlikely event that a problem relating to it is found, please inform the Central Secretariat at the address given below.

©  ISO/IEC 2010
All rights reserved. Unless otherwise specified, no part of this publication may be reproduced or utilized in any form or by any means,
electronic or mechanical, including photocopying and microfilm, without permission in writing from either ISO at the address below or
ISO's member body in the country of the requester.
ISO copyright office
Case postale 56 • CH-1211 Geneva 20
Tel. + 41 22 749 01 11
Fax + 41 22 749 09 47
E-mail copyright@iso.org
Web www.iso.org
Published in Switzerland
ii © ISO/IEC 2010 – All rights reserved

Contents Page
Foreword .v
Introduction.vi
1 Scope.1
2 Conformance .1
3 Normative references.1
4 Terms and definitions .1
5 Symbols and abbreviated terms .2
6 Fusion information format (FIF).3
6.1 Overview.3
6.2 Byte ordering .4
6.3 Numeric values .4
6.4 Fusion header block.4
7 Common elements .8
7.1 General .8
7.2 Parameter kind.8
7.3 Parameter origin .9
7.4 Distributions present .9
7.5 Number of comparisons .9
7.6 Pre-normalization flag.9
8 Type 1 record .10
8.1 Purpose .10
8.2 Format.10
8.3 Use case (Informative) .11
9 Type 2 record .12
9.1 Purpose .12
9.2 Format.12
9.3 Use case (Informative) .13
10 Type 3 record .13
10.1 Purpose .13
10.2 Format.14
Annex A (informative) Document Overview .16
Annex B (informative) Example Cumulative Distribution Functions.18
Annex C (informative) Use of pre-normalized data.20
Annex D (informative) Source for evaluation of spline .22
Bibliography.23

Figures
Figure 1 — Schematic representation of fusion information format usage.vii
Figure B.1 — Example CDFs and their spline representations .19
Figure C.1 — Example CDFs of internal comparison scores and pre-normalized scores .20
© ISO/IEC 2010 – All rights reserved iii

Tables
Table 1 — Fusion information format record structure.3
Table 2 — Fusion header block structure .3
Table 3 — Type 1 record structure.3
Table 4 — Type 2 record structure.4
Table 5 — Type 3 record structure.4
Table 6 — Textual representation of numerical value .4
Table 7 — The fusion header block .5
Table 8 — CBEFF Product Identifiers.6
Table 9 — Database identifiers.6
Table 10 — Database quality values.7
Table 11 — Score sense codes .8
Table 12 — Identifiers for statistical quantities .8
Table 13 — Origins of statistical data.9
Table 14 — Distribution information present.9
Table 15 — Pre-normalization codes.10
Table 16 — Subtype A format.10
Table 17 — Type 1 record format.10
Table 18 — Subtype B format.12
Table 19 — Type 2 record format.12
Table 20 — Subtype C format.14
Table 21 — Type 3 record format.14
Table A.1 — Fusion information format type taxonomy.16

iv © ISO/IEC 2010 – All rights reserved

Foreword
ISO (the International Organization for Standardization) is a worldwide federation of national standards bodies
(ISO member bodies). The work of preparing International Standards is normally carried out through ISO
technical committees. Each member body interested in a subject for which a technical committee has been
established has the right to be represented on that committee. International organizations, governmental and
non-governmental, in liaison with ISO, also take part in the work. ISO collaborates closely with the
International Electrotechnical Commission (IEC) on all matters of electrotechnical standardization.
International Standards are drafted in accordance with the rules given in the ISO/IEC Directives, Part 2.
The main task of technical committees is to prepare International Standards. Draft International Standards
adopted by the technical committees are circulated to the member bodies for voting. Publication as an
International Standard requires approval by at least 75 % of the member bodies casting a vote.
Attention is drawn to the possibility that some of the elements of this document may be the subject of patent
rights. ISO shall not be held responsible for identifying any or all such patent rights.
ISO/IEC 29159-1 was prepared by Joint Technical Committee ISO/IEC JTC 1, Information technology,
Subcommittee SC 37, Biometrics.
ISO/IEC 29159 consists of the following parts, under the general title Information technology — Biometric
calibration, augmentation and fusion data:
⎯ Fusion information format
© ISO/IEC 2010 – All rights reserved v

Introduction
Biometric systems embed disparate technologies and comparison algorithms. Although some of these have
been published, most are entirely proprietary. Most current verification or identification applications employ a
single biometric modality. That is, information is acquired from a body part or an exhibited behavior with the
intent of more or less uniquely identifying the individual. For example, an access control system can image the
hand and use geometrical features. A social benefits program can collect fingerprints from applicants as input
to a one-to-many duplicate search. Different biometric modes offer varying amounts of discriminative
information and have different acquisition related problems. The effect is that biometric systems are to some
extent fallible and, moreover, they exhibit different failure modes. This affords opportunities to combine
technologies or algorithms to improve performance and/or usability. Such combination is known as fusion.
Fusion can be multi-modal (e.g. observing the biometric characteristics, face and finger), multi-algorithmic (e.g.
face recognition algorithms A and B), multi-instance (e.g. index finger and thumb), multi-sensorial (e.g. optical
and ultrasound fingerprint sensor) or multi-presentation (e.g. three images of a user's face).
This part of ISO/IEC 29159 addresses the most common and most readily implemented method of fusion:
score-level fusion. This is implemented after two or more systems have processed and matched an
individual's biometric information to one or more enrolled samples and produced scalar comparison scores as
output. The scores can be either genuine (same-person) or impostor (different-person) scores and a fusion
scheme is designed to combine such scores so that the class boundary between genuine and impostor scores
is refined.
Distributions of comparison scores are unique to each biometric comparison subsystem. Score ranges and
the shapes of the distributions can differ greatly. Fusion is often implemented in two ways.
⎯ In classification-based processes, the available comparison scores are combined directly to produce an
output decision or score.
⎯ In normalization-based processes, fusion is preceded by a transformation of each score to a common
domain. Simple normalization techniques based on statistical parameters such as the mean and standard
deviation are sometimes effective, but more sophisticated techniques utilize detailed knowledge of the
entire score distribution. The fusion information format (FIF) defined in this part of ISO/IEC 29159 is
intended to flexibly support any of the popular transformations. By establishing a standardized means of
data exchange, this part of ISO/IEC 29159 supports a modular approach to biometric systems integration
in which both the comparison and fusion algorithms remain protected as black-box pieces of intellectual
property. Thus this part of ISO/IEC 29159 envisages an application in which two (or more) underlying
acquisition and comparison technologies (hand geometry and fingerprint, for example) each generate a
score which is fed to a fusion module which has been initialized with an appropriate instance of the FIF
defined herein.
Figure 1 depicts the logical role of the records in a (notional) multimodal fusion process.
This part of ISO/IEC 29159 defines containers for the distributional score information from a comparison
subsystem. It does not allow for joint distributional data that can fully capture the statistical properties of
multivariate scores (i.e. those from two or more vendors' subsystems or modalities). This means that
multimodal fusion is not supported by a description of the joint distributions of the biometric scores. This is
often a minor limitation because different modalities are often assumed to be independent. Even when the
scores are not independent, as is the case for multi-algorithm applications, score-level fusion techniques often
remain effective, even if they are not optimal.
This part of ISO/IEC 29159 is intended to support interoperability and data interchange among biometrics
applications and systems. As such it specifies requirements that solve the complexities of applying biometrics
to a wide variety of personal recognition applications, whether such applications operate in an open systems
environment or consist of a single, closed system. Open systems are built on standards based, publicly
defined data formats, interfaces, and protocols to facilitate data interchange and interoperability with other
vi © ISO/IEC 2010 – All rights reserved

systems, which can include components of different design or manufacture. A closed system can also be built
on publicly defined standards, and can include components of different design or manufacture, but inherently
has no requirement for data interchange and interoperability with any other system.
Biometric data interchange format standards and biometric interface standards are both necessary to achieve
full data interchange and interoperability for biometric recognition in an open systems environment. The
biometric International Standards developed within JTC 1/SC 37 form a layered set of International Standards
consisting of biometric data interchange formats and biometric interfaces, as well as application profiles that
describe the use of these International Standards in specific application areas.
Decision
Logic
FIF Record for FIF Record for Finger
Fusion
Face Comparison Comparison
module
Subsystem Subsystem
Initialization Initialization
Biometric Biometric
Score Score
reference reference
(image or (image or
template) template)
generated at generated at
enrolment enrolment
Face Finger
comparison comparison
subsystem subsystem
Biometric
Biometric
probe (image
probe (image
or template)
or template)
generated for
generated for
verification or
verification or
identification
identification
Component or data Component or data Component or data
supplied by vendor A supplied by vendor C supplied by vendor B

Figure 1 — Schematic representation of fusion information format usage

© ISO/IEC 2010 – All rights reserved vii

INTERNATIONAL STANDARD ISO/IEC 29159-1:2010(E)

Information technology — Biometric calibration, augmentation
and fusion data —
Part 1:
Fusion information format
1 Scope
This part of ISO/IEC 29159 specifies a biometric fusion information format that establishes machine readable
data formats to describe the statistics of comparison score inputs to a fusion process.
This part of ISO/IEC 29159 does not
⎯ standardize comparison-score normalization processes, nor
⎯ standardize or define fusion processes.
2 Conformance
Records are conformant to this part of ISO/IEC 29159 if they conform to all normative requirements of
Clause 6. This requires conformance to either Clause 8, 9, or 10, each of which requires conformance to the
stated subclauses of Clause 7.
3 Normative references
The following referenced documents are indispensable for the application of this document. For dated
references, only the edition cited applies. For undated references, the latest edition of the referenced
document (including any amendments) applies.
IEEE 754-2008, IEEE Standard for Floating-Point Arithmetic
ISO/IEC 19785-1:2006, Information technology — Common Biometric Exchange Formats Framework —
Part 1: Data element specification
ISO/IEC 19794-1:2006, Information technology — Biometric data interchange formats — Part 1: Framework
4 Terms and definitions
For the purposes of this document, the terms and definitions given in ISO/IEC 19794-1 and the following
apply.
4.1
biometric sample
analog or digital representation of biometric characteristics prior to biometric feature extraction
NOTE A biometric capture device is a biometric capture subsystem with a single component.
© ISO/IEC 2010 – All rights reserved 1

4.2
cumulative distribution function
probability that a variate takes on a value less than or equal to a number
4.3
genuine score
comparison score from comparison of two samples from one person
4.4
impostor score
comparison score from comparison of two samples from different persons
4.5
location parameter
generic measure of the position of a distribution
NOTE The location parameter is not necessarily the mean of a distribution.
4.6
probability density function
derivative of the cumulative distribution function
4.7
scale parameter
generic measure of the breadth of a distribution
NOTE The scale parameter is generally not the variance (nor the standard deviation) of a distribution.
4.8
comparison score
scalar output from biometric comparison subsystem
NOTE The term comparison score is used generically in this part of ISO/IEC 29159 for both distances (smaller
indicates greater likelihood that samples come from same person) and similarity scores (higher indicates same-person).
This distinction is conveyed explicitly in the score sense type defined in 6.4.9.
5 Symbols and abbreviated terms
For the purposes of this document, the following abbreviations apply. In tables that define binary record
structures the use of the symbol "M" in the status column indicates that the field itself is mandatory. A value of
"O" indicates the field is optional. This means that the bytes for the fields might not be present. In all cases
records can be parsed because the presence or absence of an optional field is recorded in a preceding field.
CBEFF Common Biometric Exchange Formats Framework
CDF cumulative distribution function
ECDF empirical cumulative distribution function
FAR false accept rate
FMR false match rate
FIF fusion information format
PDF probability density function
ID identifier
2 © ISO/IEC 2010 – All rights reserved

6 Fusion information format (FIF)
6.1 Overview
6.1.1 Record structure
The FIF record is used to support modularity in multi-modal biometric and decision support systems. Its format
is given in Table 1.
NOTE An application should establish a profile of part of the ISO/IEC 29159 standard. A default profile would
explicitly call out one of the record types.
Table 1 — Fusion information format record structure
Type 1 Record and/ Type 2 Record and/o Type 3 Record
Fusion header block
(see clause 6.1.3) or (see clause 6.1.4) r (see clause 6.1.5)
6.1.2 Header structure
The fusion header block structure defines the format of the record, and indicates the content. Its format is
given in Table 2.
Table 2 — Fusion header block structure
Fusion = Format Version Record Biometric
header Identifier Number Length Type
block
clause 6.4 6.4.2 6.4.3 6.4.4 6.4.5
25 bytes 4 4 4 3
Comparison Database Enrolment Verification Score Number of
Subsystem ID Database Database Sense Type
Product ID quality quality Instances
row continuation
6.4.6 6.4.7 6.4.8 6.4.8 6.4.9 6.4.10
4 2 1 1 1 1
6.1.3 Type 1 record structure
The fusion header block structure defines the format of the record, and indicates the content. Its format is
given in Table 3.
Table 3 — Type 1 record structure
Type 1 Record Type Distributions Impostor Distribution Genuine Distribution
=
Structure Present
Num Comp Loc Scale Num Comp Loc Scale
clause 8.2 8.2.3 8.2.4 8.2.5 8.2.6 8.2.7 8.2.8 8.2.9 8.2.10

26 or 50 bytes 1 1 4 10 10 4 10 10
6.1.4 Type 2 record structure
The fusion header block structure defines the format of the record, and indicates the content. Its format is
given in Table 4.
© ISO/IEC 2010 – All rights reserved 3

Table 4 — Type 2 record structure
Type 2 Record Structure Type Distributions Present Impostor Distribution CDF Genuine Distribution
=
CDF
clause 9.2 9.2.3 9.2.4 9.2.5 9.2.6

16N+13 or 32N+22 bytes 1 1 16N+11 16N+11
6.1.5 Type 3 record structure
The fusion header block structure defines the format of the record, and indicates the content. Its format is
given in Table 5.
Table 5 — Type 3 record structure
Type 3 Record Structure Type Distributions Present Impostor Distribution CDF Genuine Distribution
=
CDF
clause 10.2 10.2.3 10.2.4 10.2.5 10.2.6

16N-18 or 32N-38 bytes 1 1 16N-20 16N-20
6.2 Byte ordering
Within the FIF record, and all well-defined data blocks therein, all multi-byte quantities shall be stored in Big
Endian format. That is, the more significant bytes of any multi-byte quantity are stored at lower addresses in
memory than the less significant bytes.
EXAMPLE For example, the value 1025 (2 to the 10th power plus one) would be stored as first byte= 00000100b
and second byte=00000001b.
6.3 Numeric values
All numeric values present in the defined Types of this part of ISO/IEC 29159 are fixed-length unsigned
integer quantities, unless specified otherwise.
All numeric values given in the text of this part of ISO/IEC 29159 are decimals, unless preceded by 0x, to
indicate hexadecimal, or suffixed by a "b" to indicate binary.
Table 6 — Textual representation of numerical value
Example value Radix Decimal value
1010b 2 10
39 10 39
0xF5 16 245
Double precision numbers shall be conformant with IEEE 754.
NOTE The specification of IEEE 754 may not be sufficient to avoid numerical inaccuracy.
6.4 Fusion header block
6.4.1 General
The fusion header block of Table 7 shall be present as the first block in all FIF records.
4 © ISO/IEC 2010 – All rights reserved

Table 7 — The fusion header block
Field Status Size (bytes) Valid Values Notes
Format identifier M 4 0x46494600 ASCII string "FIF" with null terminator
Version number M 4 0x30313000 ASCII string "010" with null terminator
Record length M 4 1 ≤ L ≤ 2 -1 Length of entire record in bytes
Biometric type M 3 0 ≤ t ≤ 0x080000 The modality from which this record came from
Comparison subsystem
M 4 ≥ 0 Currently vendor specified ID.
product ID
Database ID M 2 ≥ 0 Currently vendor specified ID.
Enrollment sample M 1 [0-100],254,255 Aggregate qualities of the samples used
quality
to compute the comparison score statistics
Verification sample
M 1 [0-100],254,255
quality
Score sense M 1 0 or 1 Distances or similarities? See clause 6.4.9.
Number of type instances M 1 1 ≤ N ≤ 4 0 is not allowed
6.4.2 Format identifier
The (4-byte) format identifier shall begin with the three ASCII characters ‘FIF’ to identify the record as
following this part of ISO/IEC 29159, followed by a zero byte as a null string terminator.
6.4.3 Version number
The (4-byte) version number shall consist of three ASCII numerals followed by a zero byte as a NULL string
terminator. The value denotes the version of this part of ISO/IEC 29159.
The first and second character represents the major version number and the third character represents the
minor revision number.
The version number shall be 0x30313000 which is “010” – Version 1 revision 0.
6.4.4 Record length
The (4-byte) record length block shall be the combined length in bytes for the entire record. It shall be the sum
of the fusion header block length (25 bytes) and any Type 1, 2 or 3 record lengths that follow.
6.4.5 Biometric type
The (3-byte) value shall be taken from CBEFF's enumeration of biometric modalities given in
ISO/IEC 19785-1:2005, 6.5.6. This value allows an application to know what biometric modality is represented
by the instance of the fusion information format.
EXAMPLE For scores from an implementation using vein patterns this value would be 0x040000.
6.4.6 Comparison subsystem Product ID
The (4-bytes = 2 + 2 bytes) identifier of the comparison subsystem algorithm (i.e. comparing the biometric
probe with the biometric reference) that produced the scores information described in this FIF record shall be
recorded according to Table 8. These two values are the CBEFF Product Identifiers described in
ISO/IEC 19785-1:2006.
© ISO/IEC 2010 – All rights reserved 5

Table 8 — CBEFF Product Identifiers
Assigned by Length Description
Product ID (stored in the first two bytes) IBIA 2 bytes See ISO/IEC 19785-1:2006
Version Number (stored in the second two bytes) Vendor 2 bytes
NOTE 1 It can be argued that a Product ID field is not needed because the statistical information embodies all that is
needed by a fusion module. The field is however useful to support, for example, version control and caching.
NOTE 2 The Product ID alone is likely insufficient to execute score normalization. It may nevertheless be useful to
applications using the records of this part of ISO/IEC 29159.
6.4.7 Database ID
The (2-byte) vendor-specified identifier of the database used by the vendor's comparison subsystem in
creation of the scores used in the computation of the data contained in fields 4 and 5 shall be recorded
according to Table 9. This field might be used
⎯ by a comparison subsystem provider to arbitrarily indicate what data they used in computing the FIF
record contents, or
⎯ in an application profile or requirements document to firmly specify that suppliers must use a specific
entry (for example item 49, the MINEX POEBVA fingerprint calibration database).
NOTE ISO/IEC JTC 1 SC 37 maintains a process for registration of other databases for use in generating
ISO/IEC 29159-1 records. The Registry of Database IDs includes the databases listed below.
Table 9 — Database identifiers
Dataset ID Distribution Modality
0 Unspecified
1 Unknown
2 Composite
3-15 Reserved
16. FERET Face Database http://www.itl.nist.gov/iad/humanid/colorferet/ Face
17. Yale Face Database http://cvc.yale.edu/projects/yalefaces/yalefaces.html Face
18. PIE Database CMU http://www.ri.cmu.edu/projects/project_418.html Face
19. AR Database http://cobweb.ecn.purdue.edu/~aleix/aleix_face_DB.html Face
20. FRGC I Database http://face.nist.gov/frvt/ Face
21. FRGC II Database http://face.nist.gov/frvt/ Face
22. FRVT 2002 HCInt (i.e. FRVT 2006 Low Resolution) Database http://face.nist.gov/frvt/ Face
23. FRVT 2006 High Resolution Database http://face.nist.gov/frvt/ Face
24. FRVT 2006 Very High Resolution Database http://face.nist.gov/frvt/ Face
25. FVC 2000 Finger
26. FVC 2002 DB1 http://bias.csr.unibo.it/fvc2002/ Finger
27. FVC 2002 DB2 http://bias.csr.unibo.it/fvc2002/ Finger
28. FVC 2002 DB3 http://bias.csr.unibo.it/fvc2002/ Finger
29. FVC 2002 DB4 (synthetic) http://bias.csr.unibo.it/fvc2002/ Finger
30. FVC 2004 DB1 http://bias.csr.unibo.it/fvc2004/ Finger
31. FVC 2004 DB2 http://bias.csr.unibo.it/fvc2004/ Finger
32. FVC 2004 DB3 http://bias.csr.unibo.it/fvc2004/ Finger
33. FVC 2004 DB4 (synthetic) http://bias.csr.unibo.it/fvc2004/ Finger
34. FVC 2006 DB1 http://bias.csr.unibo.it/fvc2006/ Finger
35. FVC 2006 DB2 http://bias.csr.unibo.it/fvc2006/ Finger
36. FVC 2006 DB3 http://bias.csr.unibo.it/fvc2006/ Finger
6 © ISO/IEC 2010 – All rights reserved

Dataset ID Distribution Modality
37. FVC 2006 DB4 (synthetic) http://bias.csr.unibo.it/fvc2006/ Finger
38. NIST Special Database 27 http://fingerprint.nist.gov Latent finger
39. NIST Special Database 29 http://fingerprint.nist.gov Finger
40. MCYT Fingerprint subcorpus http://atvs.ii.uam.es/bbdd_EN.html Finger
41. MCYT Signature subcorpus http://atvs.ii.uam.es/bbdd_EN.html Signature
42. BANCA still http://www.ee.surrey.ac.uk/banca/ Face
43. BANCA video http://www.ee.surrey.ac.uk/banca/ Voice
44. BANCA high quality http://www.ee.surrey.ac.uk/banca/ Voice
45. BANCA low quality http://www.ee.surrey.ac.uk/banca/ Voice
46. NIST Speaker Verification http://www.nist.gov/speech/tests/spk/2005/ Voice
47. NIST Speaker Verification http://www.nist.gov/speech/tests/spk/2006/ Voice
48. CASIA Iris http://www.nlpr.ia.ac.cn/english/irds/irisdatabase.htm Iris
49. Bath Iris http://www.irisbase.com Iris
50. ICE 2005 http://iris.nist.gov/ice/ Iris
51. ICE 2006 http://iris.nist.gov/ice/ Iris
52. NIST MINEX DHS2 calibration set http://fingerprint.nist.gov Finger
53. NIST MINEX POE calibration set http://fingerprint.nist.gov Finger
54. NIST MINEX DOS calibration set http://fingerprint.nist.gov Finger
55. NIST MINEX POEBVA calibration set http://fingerprint.nist.gov Finger
56. TURBINE GUC100 http://www.nislab.no/guc100 Finger
57.-2047 Reserved
2048 - 65535 Vendor assigned, possibly non-public and non-unique across vendors
NOTE 1 Although some of these databases are not available to the public they are listed here to support future
calibration efforts.
NOTE 2 Although some of these databases are not freely available, they are listed here to point out their existence to
users of this part of ISO/IEC 29159.
NOTE 3 The performance of fusion systems will have some sensitivity to the databases used. This is because the
comparison score distributions obtained by matching samples from one database will generally be different than from
another. See Annex C for a discussion of the stability of distributions.
6.4.8 Database quality
The qualities of the samples in a database may have been aggregated together to form a scalar database
quality value. Two (1-byte) quality summary values shall be stored in consecutive fields, the first for the
reference samples, the second for the probe samples. A value of 0 shall represent the lowest possible quality
and the value 100 shall represent the higher possible quality. For each field, the allowed values are given in
Table 10.
Table 10 — Database quality values
Value Meaning
0 - 100 Assigned value
254 Unassigned because no attempt was made to compute value
255 Unassigned because of failed attempt to assign a quality
score
6.4.9 Score sense
The raw outputs from biometric comparison subsystems may be either distances or similarities. Small values
for distances indicate greater likelihood that the score is a genuine score; for similarities, higher values
indicate this. A 1 byte field with values allowed by Table 11 shall be used to record score sense.
© ISO/IEC 2010 – All rights reserved 7

Table 11 — Score sense codes
Sense Value (1 byte)
Distance 0
Similarity 1
NOTE The use of the word distance does not necessarily imply that values have the metric property.
6.4.10 Number of type instances
The (1-byte) number of type instances block shall be the number of Type 1, 2, or 3 records included in the FIF
record. A record shall contain zero or one instance of each Type, but there shall always be one instance of
one of them. Thus the number of type instances shall be on 1, 2 or 3.
EXAMPLE If both a Type 1 and Type 3 instance are present, this value will be 2.
NOTE An application profile (or equivalent specification) might appropriately call-out one specific type.
7 Common elements
7.1 General
This clause defines generic data structures to support the Type 1, 2 and 3 records of this part of
ISO/IEC 29159. This clause also includes tables that enumerate values and associates them with their
respective meanings.
Note that in some tables the first column is entitled "Field" and contains a numeric entry. These are included
only for ease of reference to the rows – the values should not be included in instances of the standard's binary
records.
7.2 Parameter kind
Table 12 gives integer values as identifiers for quantities needed to describe a distribution.
Table 12 — Identifiers for statistical quantities
Kind Description Notes
0. unspecified the producer neglects to specify (usually unacceptable)
1. unknown unknown quantity
2. mean
3. median
4. mode
5. minimum extreme values not true location parameters
6. maximum
7. (minimum + maximum) / 2
8. Tukey's trimean
9. generic location parameter
10. – 31. undefined
32. variance
33. standard deviation
Actually 1.4826 (med(abs(x-med)) where the constant 1.4826
34. median absolute deviation (MAD)
normalizes so that the expectation value of the mad is equal to the std.
deviation of a normal distribution.
35. (maximum – minimum) extreme values not true location parameters
36. generic scale parameter
37. skewness
8 © ISO/IEC 2010 – All rights reserved

Kind Description Notes
38. kurtosis
39. – 65. undefined
66. generic parameter I for distributions parameterized by more than just location and scale
67. generic parameter II
68. – 95. undefined
96. cumulative distribution function as discrete pairs (x , F(x)) for i = 1…N as used in Type 2
i i
97. cumulative distribution function as a B-spline representation used in Type 3
98. – 255 undefined
NOTE 1 An application may well effect the same transformation regardless of which kind of values the record holds,
e.g. linear transforms such as (x-mean)/stdev and (x-median)/mad.
NOTE 2 The range of parameters is wide. For a specific application it will be necessary to narrowly specify these. This
is properly done in an application profile, or other requirements document.
7.3 Parameter origin
The (1 byte) value of Table 13 shall be used to indicate the origin of the parameter.
Table 13 — Origins of statistical data
Origin Distribution of what quantity
0. Unspecified – value(s) whose origin is not divulged
1. Unknown – value(s) with undetermined origin
2. Empirical – value(s) estimated from experiment and sample data
3. Known a priori – value(s) known by design, or theoretical considerations. For a
distribution this value would indicate it is known in closed-form.
4. – 254 Reserved
7.4 Distributions present
The (1 byte) value shall be used to indicate which of the impostor and genuine distributions are included in a
record. The allowed values are given in Table 14. This field shall contain the value 0x01, 0x02 or 0x03. When
both distributions are present the impostor distribution shall precede the genuine distribution.
Table 14 — Distribution information present
Value Distributions present
0x01 Impostor
0x02 Genuine
0x03 Impostor and Genuine
0x04 - 0xFF Reserved
7.5 Number of comparisons
The (4-byte) number of genuine or impostor comparisons used to estimate the Type 1, 2 or 3 information shall
be encoded in this field. A value of zero shall be used when this value is unknown.
7.6 Pre-normalization flag
The (1 byte) value of Table 15 shall be used to indicate whether scores from the comparison subsystem exist
on an arbitrary range or have been pre-normalized. For the purposes of this part of ISO/IEC 29159, this
means that a fusion module may regard the scores produced by the comparison subsystem to be uniformly
© ISO/IEC 2010 – All rights reserved 9

distributed on the range [0,1]. This supports direct use of scores without need for interpolation. See Annex C
for information on the meaning of this field.
NOTE 1 This field is present in the Type 2, or 3 records for the genuine and impostor distributions. If a comparison
subsystem produces pre-normalized impostor scores, the genuine scores will not be uniform, and vice versa. Thus it will
never make sense for the records of this part of ISO/IEC 29159 to indicate that both the impostor and genuine scores are
pre-normalized.
NOTE 2 Implementations conformant to BioAPI [1] return "FAR values". This means that internally generated impostor
scores are either naturally on [0,1] or have been normalized by the expected impostor CDF of those scores. The result will
be that the output impostor scores can be expected to be uniform on [0,1].
Table 15 — Pre-normalization codes
Pre-normalization status Range of scores Value (1 byte)
Not normalized -∞ < x < ∞ 0
Pre-normalized 0 ≤ x ≤ 1 1
8 Type 1 record
8.1 Purpose
The Type 1 record holds minimal statistical information about the impostor and/or genuine scores from a
biometric system. This information may be used for rescaling scores before a fusion operation.
8.2 Format
8.2.1 Supporting datatype - Subtype A
Subtype A contains the kind (see clause 7.2), origin (clause 7.3) and value of a scalar statistic. The format is
given in Table 16.
Table 16 — Subtype A format
Field Status Description Length (bytes) Data type Allowed values Example
1 M Parameter kind 1 uint See Table 12 3 (median)
2 M Parameter origin 1 uint 0-3 2 (empirical)
3 M Parameter value 8 double Range of double 2.998 x 10
8.2.2 Definition
The format of a Type 1 record shall be that shown in Table 17. Either the impostor data or the genuine data or
both shall be present (i.e. the valid sizes are 26 or 50 bytes).
Table 17 — Type 1 record format
Field Status Description Length (bytes) Data type Allowed values Example
1. M Type 1 uint 1 1
2. M Distributions present 1 packed bit field 0x01, 0x02, 0x03 0x01
3. O Impostor number of comparisons 4 uint [0,2 - 1] 40000
4. Impostor distribution location 10 Subtype A ( 3 1 2.998)
5. Impostor distribution scale 10 Subtype A (34 1 0.308)
6. O Genuine number of comparisons 4 uint [0,2 - 1] 240
7. Genuine distribution location 10 Subtype A ( 3 1 8.310)
8. Genuine distribution scale 10 Subtype A (34 1 1.406)
10 © ISO/IEC 2010 – All rights reserved

8.2.3 Type
The Type for a Type 1 record shall be 1 stored in 1 byte.
8.2.4 Distributions present
Those distributions present in the Type 1 record shall be recorded in the bits of a Distributions Present field.
8.2.5 Impostor number of comparisons
The number of comparisons used in the computation of the impostor score statistics shall be recorded as a
Number of Comparisons field.
8.2.6 Impostor distribution location
If present, this Subtype A field shall contain the kind, origin and value for a location parameter of the impostor
score distribution. If present, this value shall be followed by the scale information in 8.2.7.
NOTE If it is sensible that a normalization scheme involves only translation (i.e. no scaling), then the scale parameter
of 8.2.7 can be set 1.
8.2.7 Impostor distribution scale
If present, this Subtype A field shall contain the kind, origin and value for a scale parameter of the impostor
score distribution.
8.2.8 Genuine number of comparisons
The number of comparisons used in the computation of the genuine score statistics shall be recorded as a
Number of Comparisons field.
8.2.9 Genuine distribution location
If present, this Subtype A field shall contain the kind, origin and value for a location parameter of the genuine
score distribution. If present, this value shall be followed by the scale information in 8.2.10.
8.2.10 Genuine distribution scale
If present, this Subtype A field shall contain the kind, origin and value for a scale parameter of the genuine
score distribution.
8.3 Use case (Informative)
Given two Type 1 records, one from a fingerprint comparison subsystem (A) and another from an iris
comparison subsystem (B), score-level fusion could be implemented as the sum of z-normalized scores as:
s = (a – amean) / asigma  +  (b – bmean) / bsigma
where
a = raw score from fingerprint comparison subsystem
b = raw score from iris comparison subsystem
amean = estimated mean of impostor scores from comparison subsystem A
estimated standard deviation of impostor scores
...