Standard

INCITS M1/07-0129
i
Revision of ANSI®
INCITS 379-2004
American National Standard
for Information Technology – Iris Image Interchange Format
Secretariat
Information Technology Industry Council (ITI)
American National Standards Institute, Inc.
Abstract
This standard describes a format for exchange of iris image information. It contains a definition of attributes, a data record format, sample records and conformance criteria. Two alternative formats for iris image data are described, one based on a cartesian coordinate system and the other on a polar coordinate system.
ii
Approval of an American National Standard requires review by ANSI that the requirements for due process, consensus, and other criteria for approval have been met by the standards developer. Consensus is established when, in the judgement of the ANSI Board of
Standards Review, substantial agreement has been reached by directly and
materially affected interests. Substantial agreement means much more than a
simple majority, but not necessarily unanimity. Consensus requires that all
views and objections be considered, and that a concerted effort be made
towards their resolution.
The use of American National Standards is completely voluntary; their
existence does not in any respect preclude anyone, whether he has approved
the standards or not, from manufacturing, marketing, purchasing, or using
products, processes, or procedures not conforming to the standards.
The American National Standards Institute does not develop standards and will
in no circumstances give an interpretation of any American National Standard.
Moreover, no person shall have the right or authority to issue an interpretation
of an American National Standard in the name of the American National
Standards Institute. Requests for interpretations should be addressed to the
secretariat or sponsor whose name appears on the title page of this standard.
CAUTION NOTICE: This American National Standard may be revised or
withdrawn at any time. The procedures of the American National Standards
Institute require that action be taken periodically to reaffirm, revise, or withdraw
this standard. Purchasers of American National Standards may receive current
information on all standards by calling or writing the American National
Standards Institute.
Published by
American National Standards Institute, Inc.
11 West 42nd Street, New York, NY 10036
Copyright © 2007 by Information Technology Industry Council (ITI)
All rights reserved.
No part of this publication may be reproduced in any
form, in an electronic retrieval system or otherwise,
without prior written permission of ITI, 1250 Eye Street NW,
Washington, DC 20005. Printed in the United States of America
American
National
Standard
INCITS M1/07-0129
iii
Contents Page Foreword............................................................................................................................................................iv Introduction.. (ix)
1 Scope (1)
2 Conformance (1)
3 Normative references (1)
4 Terms, definitions, symbols, units and abbreviated terms (2)
5 Iris image format specification (3)
5.1 General (3)
5.2 Image compression (3)
5.3 Image
5.4 Iris
image
5.5 Iris 5.6 Data values Annex A (informative) Iris image capture best practice A.1 Image quality (13)
A.1.1 Optical and Digital Resolution (13)
A.1.2 Focus Quality (13)
A.1.3 Image Quality (13)
A.2 Grayscale density (14)
A.3 Illumination (14)
A.4 Contrast (14)
A.5 Visible iris (15)
A.6 Pixel aspect ratio (15)
A.7 Image scale (15)
A.8 Optical distortion (16)
A.9 Noise (16)
A.10 Image orientation (16)
A.11 Presentation (16)
Annex B (informative) Sample iris image data records (17)
B.1 B.2 Rectilinear data record sample, multiple eyes, multiple images B.3 Bibliography
Foreword
INCITS (The International Committee for Information Technology Standards) is the ANSI recognized Standards Development Organization for information technology within the United States of America. Members of INCITS are drawn from Government, Corporations, Academia and other organizations with a material interest in the work of INCITS and its Technical Committees. INCITS does not restrict membership and attracts participants in its technical work from 13 different countries, and operates under the rules of the American National Standards Institute.
In the field of Biometrics, INICTS has established the Technical Committee M1. Standards developed by this Technical Committee have reached consensus throughout the development process and have been thoroughly reviewed through several Public Review processes.
At the time that it approved this standard, INCITS had the following members:
iv
INCITS M1/07-0129
v
Technical Committe M1 on Biometrics, which reviewed this standard, had the following members:
Fernando Podio, Chair
Colin Soutar, Vice-Chair
Organization Represented
Name of Representative P – Principal A – Alternate 3M CO
Campbell, John (P)
Larson, Roger (A) Assa Abloy ITG
Pefley, Marian (P)
Quan, Ralph (A) AuthenTec, Inc.
Setlak, Dale (P) Authenti-Corp
Valencia, Valorie (P) AWARE INC
Benini, David (P)
Mungovan, Rob (A) Bioscrypt, Inc.
Soutar, Colin (P) Business Solutions
Eul, John (P) Cross Match Technologies, Inc.
Ruggles, Thomas (P) Cannon, Greg (A) DCTA
Thewlis, Dave (P) DataCard Group
Baggeroer, Chuck (P) IBIA
Richard Norton (P) ID Technology Partners, Inc.
Jerde, Mark (P)
Herr, Fred (A) Identix Corporation
Rudolph Nobel, Kirsten (P)
Griffin, Paul (A)
Bowman, Erik (A) INFINEON TECHNOLOGIES
Stafford, Mark (P)
Majid, Ashi (A) International Biometric Group
Mak, Mken (P)
Thieme, Mike (A) Iridian Technologies
Cambier, James (P) Iritech, Inc.
Kim, Daehoon (P) LASERCARD SYSTEMS CORP
Price-Francis, Stephen (P) LOCKHEED MARTIN CORP
Sutton, Ronald (P)
Florsek, Donald (A) Mastercard
Russell, James (P)
Nardone, Christopher (A) Mississippi Valley State University
Gilligan, Allan (P) MITRETEK SYSTEMS
D'Amato, Donald (P)
Kiebuzinski, George (A) NATIONAL SECURITY AGENCY
Dunn, Jeffrey (P)
Reinert, Lawrence (A)
King, Matthew (A) NEC Solutions
Giordano, Theresa (P)
Gousie, Stephen (A) NIST Podio, Fernando (P)
Hogan, Michael (A)
McCabe, Michael (A) OSS NOKALVA Triglia, Alessandro (P)
Scott, Bancroft (A) Purdue University Elliott, Stephen (P) Q.E.D. Systems Harmon, Craig (P) SAFLINK Corp. Tilton, Cathy (P) SAGEM MORPHO INC Jones, Creed (P) Security Industry Association Visbal, Mark (P) SONY ELECTRONICS INC Harris, John (P)
Barrett, Ed (A)
ST MICROELECTRONICS Fabbrizio, Vito (P) Transaction Security, Inc. Beatson, Rod (P) Unisys Corp Schaffner, Ed (P)
Vinsik, Steve (A) United States Dept. of Defense - BMO/BFC Hapeman, Dale (P) United States Dept. of Homeland Security Lazarick, Rick (P)
Neumann, John (A)
Ruwaldt, Paul (A) United States Dept. of Justice Miles, Christopher (P)
Gilchrist, Mike (A) United States Dept. of State Kefauver, Barry (P)
Williams, Curtis (A) West Virginia High Technology Consortium Foundation Gaydos, Kenneth (P)
Malnick, Kathy (A)
vi
INCITS M1/07-0129
vii
Task Group M1.1 on Biometric Data Interchange Formats, which developed this standard, had the following participants :
Creed Jones, Chair
James Cambier, Vice-Chair and Iris Interchange Technical Editor
Organization Represented
Name of Representative P – Principal A – Alternate 3M Co
Campbell, John (P)
Larson, Roger (A) Atmel
Mainguet, Jean-Francois (P)
Ruffin, Philippe (A) AuthenTec, Inc.
Setlak, Dale (P) Aware Inc
Benini, David (P)
Maurer, Jim (A) Bioscrypt, Inc.
Soutar, Colin (P) CrossMatch Cannon, Greg (P) Identix Corporation
Nobel, Kirsten (P)
Griffin, Paul (A)
Bowman, Erik (A) Iridian Technologies
Cambier, James (P) Iritech, Inc.
Kim, Daehoon (P) Lockheed Martin Corp
Sutton, Ronald (P) Mitretek Systems
D'Amato, Donald (P) Motorola Wabgaonkar, Harshawardhan (P)
Hall, David (A) National Biometric Security Project
Williams, Gerald (P) National Security Agency
Dunn, Jeffrey (P)
King, Matthew (A) NEC Solutions
Giordano, Theresa (P)
Gousie, Stephen (A) NIST
McCabe, Michael (P)
Hogan, Michael (P) OSS Nokalva
Triglia, Alessandro (P)
Scott, Bancroft (A) Passports Australia
Hartmann, Terry (P)
Osborne, John (A) Recognition Systems, Inc.
Tamer, Samir (P)
Gaughan, Chris (A) SAFLINK Corp.
Tilton, Cathy (P) Sagem Morpho Inc
Jones, Creed (P) Fineman, Dan (A) Sony Electronics Inc
Harris, John (P)
Barrett, Ed (A) The Aerospace Corporation
Buettner, Doug (P) Transaction Security, Inc.
Beatson, Rod (P) Unisys Corp
Vinsik, Steve (P)
Organization Represented Name of Representative
P – Principal
A – Alternate
Schaffner, Ed (A)
United States Dept. of Defense – BMO/BFC Guirguis, Ramy (P)
Randall, Rick (A)
United States Dept. of Defense – DISA Walker, Ric (P)
United States Dept. of Homeland Security Lazarick, Rick (P)
Neumann, John (A)
Wing, Brad (A)
United States Dept. of State Kefauver, Barry (P)
Williams, Curtis (A)
viii
INCITS M1/07-0129
Introduction
The purpose of this document is to define a proposed standard for exchange of iris image information. The standard contains a specific definition of attributes, a data record format for storing and transmitting the iris image and certain attributes, several sample records, and conformance criteria.
Currently, exchange of iris information between equipment from different vendors can only be done using a large-scale image of the entire eye. This is expensive in storage and bandwidth. To provide interoperability among vendors, it is necessary to define a standard, compact representation of a human iris.
ix
M1/03-0167
Iris Image Interchange Format
1 Scope
This Standard specifies two alternative image interchange formats for biometric authentication systems that utilize iris recognition. The first is based on a rectilinear image storage format that may be a raw, uncompressed array of intensity values or a compressed format such as that specified by the JPEG standard. Images may be monochrome or color with 256 or more intensity levels (grey or per-color), and vary in size depending on field of view and compression. Typical size is 25 –30 Kbytes for JPEG format.
The second format is based on a polar image specification that requires certain pre-processing and image segmentation steps, but produces a much more compact data structure that contains only iris information. The record size can be as small as 2 Kbytes. The polar image may be either raw or compressed format.
Data that comply with either one of the iris image formats specified in this standard are intended to be embedded in a CBEFF-compliant structure in the CBEFF Biometric Data Block (BDB).
2 Conformance
Conformity with the standard requires compliance with one of the data formats described in Section 5. Recommended image quality criteria described in Annex A are expressed in terms of resolution, contrast, noise level, etc. Quality is specified as a value from 0 to 100, where value 0 means “unspecified”. In general, the highest image quality levels are recommended for high-volume, high-security applications where the lowest possible recognition error rates are required. Lower levels of image quality are appropriate for less demanding applications in which higher error rates can be tolerated but camera cost is a critical factor.
This standard is intended for technical review of the proposed image standard.
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.
[CBEFF] NISTIR 6529-A – Common biometric exchange file format - augmented
[BioAPI] ANSI/NIST 358-2002 – Information technology – BioAPI specification
[X9.84] X9.84-2001 – Biometric information management and security
[JPEG] ISO/IEC 10918-1:1994, ISO/IEC 10918-2:1995, ISO/IEC 10918-3:1997, ISO/IEC 10918-4:1999 – Information technology – Digital compression and coding of continuous-tone still images (JPEG)
[JPEG2000] ISO/IEC 15444-1:2000, ISO/IEC 15444-3:2002, ISO/IEC 15444-4:2002 – Information technology – JPEG 2000 image coding system
4 Terms, definitions, symbols, units and abbreviated terms
For the purposes of this document, the following terms, definitions, units and abbreviated terms apply.
4.1
BLOB
binary large object, a large block of binary data, typically an image, that may have to be handled in a special way
4.2
continuous tone image
image whose components have more than one bit per sample
4.3
crop
to reduce the size of an image by removing unimportant parts of it, typically the outer regions
4.4
grey scale
continuous-tone image that has one component, which is luminance
4.5
iris
colored annular ring in the front portion of the eye comprised of muscular and connective tissue and pigment cells that defines and controls the central opening of the pupil
4.6
JPEG
compression standard for continuous-tone images, published in 1994 as ISO 10918-1 and ITU-TT.81
4.7
JPEG2000
enhanced compression standard published as ISO/IEC 15444
4.8
limbus
outer boundary of the iris where it is joined to the sclera
4.9
lp
line pair, a measure of spatial feature content, which when associated with a fixed distance measurement provides an estimate of spatial frequency
4.10
mm
millimetre
4.11
pixel
single picture element, one of an m by n matrix of picture elements where m is the number of columns and n is the number of rows
4.12
pupil
opening in the center of the eye that serves as a variable light aperture and defines the inner boundary of the iris
INCITS M1/07-0129 4.13
raw
image file format in which the image is stored in the same format in which it is stored in video memory, typically one byte (for monochrome images) per picture element or three bytes (for color images) per picture element
4.14
resolution
number of picture elements (pixels) per unit distance in the object plane or image plan, specified as the number of pixels per millimetre at the object plane, that is on the eye
4.15
round
mathematical function applied to a number x such that round(x) is the integer that is closest in value to x
4.16
sclera
white outer covering of the eye peripheral to the iris
5 Iris image format specification
5.1 General
The iris image format specification defines header and data structures that support storage of the iris image in rectilinear or polar coordinates.
The biometric data record specified in this standard shall be embedded in a CBEFF-compliant structure in the CBEFF Biometric Data Block (BDB). The CBEFF Format Owner assigned by the International Biometric Industry Association (IBIA) to INCITS Technical Committee M1 shall be used. This is the sixteen-bit value 0x001B (hexadecimal 1B or 27 decimal). It is recommended that iris image data transmitted or stored in accordance with the standard defined here be encrypted and signed to protect privacy and integrity of the data. The CBEFF structure supports such encryption and signing.
One of the two following CBEFF Format Type values shall be used in the CBEFF Header. The sixteen-bit value 0x0601 shall be used for records that represent the iris image in rectilinear coordinates and the sixteen-bit value 0x0602 shall be used for records that represent the iris image in polar coordinates.
5.2 Image compression
5.2.1 General - The iris image shall be transmitted and stored in one of several possible formats described in the following sub-clauses.
5.2.2 Raw format – The image shall be represented as an array of m columns by n rows by at least 8 bits. There is no image header, and each pixel in a monochrome image shall be represented by eight or more bits. Color images shall be represented as three samples per pixel, each comprised of eight or more bits, representing red, blue, and green intensities, in that order. The image shall be organized in row-major order, with the lowest address corresponding to the upper left corner of the image.
5.2.3 Lossless compression format -If lossless compression is used, the image data shall be compressed in accordance with JPEG2000 lossless compression specified in ISO/IEC 15444.
5.2.4 Compressed format – If lossy compression is used the image shall be compressed in accordance with the JPEG compression algorithm specified in ISO/IEC 10918 or in accordance with JPEG2000 lossy compression specified in ISO/IEC 15444.
5.3 Image pre-processing
5.3.1 Rectilinear image pre-processing - If the image is collected by a camera that captures only one eye at a time and stored using a rectilinear coordinate system no specific pre-processing is required. Cameras that capture images of both eyes simultaneously may use the following processing steps to calculate the rotation angle of the iris images.
5.3.1.1 Rectilinear image rotation angle – Iris image capture systems that record images of both eyes simultaneously may have the capability to measure the angle of the subject’s head by defining a line between the pupil centers of the left and right eyes and determining the angular difference between this line and the horizontal axis of the imaging system. If this rotation is measured and recorded, it shall be recorded as the angle in degrees between the horizontal axis of the camera system and the line between the pupil centers of the two eyes, with a positive value signifying counter-clockwise rotation of the inter-pupil line relative to the camera’s horizontal axis.
5.3.1.2 Rectilinear image rotation uncertainty - The rotation uncertainty is an estimate, dependent on the imaging device, of the maximum rotation error associated with the rotation angle. It shall be recorded as a positive nonzero value measured in degrees. If rotation information is not available the rotation uncertainty value shall be set to the maximum possible value.
5.3.2 Polar image pre-processing - If the polar coordinate system is used the following pre-processing operations shall be performed on the rectilinear image in order to convert it to polar form.
5.3.2.1 Boundary extraction – The boundaries of the pupil and iris may be extracted using a variety of techniques and either parametric or nonparametric representations. The polar image consists of a number of samples along radial lines positioned at regular angular intervals. The position of each pupil and outer iris boundary point along its radial line shall be determined with a precision of plus or minus one pixel. The polar image contains iris data between the inner boundary with the pupil and the outer boundary with the sclera.
5.3.2.2 No boundary extraction - An alternative storage format is provided, in which the estimated pupil center is used as the inner boundary and the polar image samples extend radially from the pupil center to a supplier-defined outer circle that encloses the entire iris. In this case the number of radial samples in the polar image shall be not less than the iris radius in the rectilinear image expressed in pixels, and the number of annular samples shall be not less than one-half the iris/sclera boundary length in the rectilinear image expressed in pixels. Users of this format are expected to apply post-processing algorithms to extract the precise inner and outer iris boundaries using suitable techniques.
5.3.2.3 Iris occlusions – Areas obscured by specular reflections, eyelids, eyelashes, etc. may be located and special intensity values assigned. If such assignment is performed such pixels shall be assigned a reserved iris occlusion value, usually maximum intensity or zero. The occlusion value shall be defined in the header. If such iris occlusion processing is performed and occluded areas are filled by a reserved value, subsequent compression applied to the image shall use only lossless compression algorithms.
5.3.2.4 Scan type – Corrections to accommodate specific scan types such as progressive or interlaced should be applied prior to conversion to polar coordinates. If such corrections are applied the scan type entry in the image properties bitfield shall be set to SCAN_TYPE_CORRECTED.
5.3.2.5 Orientation correction – Transformations to correct for horizontal or vertical flipping of the image shall be applied prior to conversion to polar coordinates, and the entries for horizontal and vertical orientation in the image properties bitfield shall be set to ORIENTATION_UNDEF or ORIENTATION_BASE.
INCITS M1/07-0129 5.3.2.6 Polar conversion - Image data between the inner and outer iris boundaries shall be converted to polar coordinates, with each pixel represented by one or more (for color) intensity values. The inner and outer boundaries, which need not be circular or concentric, shall be each divided into m angular intervals. The image segment extending from the i th inner boundary segment to the i th outer boundary segment shall be divided into n radial samples. The intensity of each polar image sample p(r,θ) shall be computed using nearest-neighbor re-sampling or other industry-standard method. The zero degree angular value shall be at the 6 o’clock position, directly downward from the estimated pupil center, and angular values shall increase in the counter-clockwise direction. The pixel with the lowest address shall be adjacent to the inner boundary at 0 degrees. Subsequent addresses shall be occupied by pixels at the same radius but increasing angles until the first ring is completed. This shall be followed by m samples at radius = 1, etc. as shown in Figure 1a (for m=256 and boundary extraction) and figure 1b (for m=256 without boundary extraction). The entire uncompressed image shall consist of n x m pixels for a monochrome image having 8 or more bits per pixel intensity information or 3 x n x m for a color image having 8 or more bits of intensity information per pixel for each color. Figure 2 is a representative iris image showing the results of pre-processing steps that extract the inner and outer iris boundaries using the circular models. The extracted pupil, iris, and eyelid boundaries are shown along with the iris center. The polar representation of this same iris is shown in Figure 3. Note that the pupil boundary is at the top and the outer
θ=0θ=1
Figure 1a — Polar image sampling grid
?=0?=1
Figure 1b --- Polar image sampling grid (no boundary extraction)
Figure 2 — Iris image with features
Figure 3 —Polar iris image
INCITS M1/07-0129 5.3.2.7 Polar image rotation correction – If the rotation angle is available when the polar image is produced, it may be used to correct for rotation, so that the 0 degree radial is perpendicular to the line between the pupil centers. If such correction is performed the rotation uncertainty shall be set to a nonzero value indicating the expected accuracy of the correction. If rotation correction is not performed the rotation uncertainty shall be set to its maximum possible value. The rotation angle field is not used for polar images, and the value indicating undefined rotation angle shall be entered.
5.4 Iris image data record
Table 1 illustrates the structure of the iris data record. Every record shall have an iris record header that contains information about the image capture device and conditions. The record shall contain images from one or two eyes, which are designated iris features, and the iris record header shall indicate how many features were recorded (one or two). Each iris feature shall have an iris feature header that designates it as a right or left eye, and contains information on the number of images recorded from that eye. If the capture device is unable to determine which eye was presented, then the eye designation shall be entered as "unknown", and all capture images shall be saved with a single feature header. Each iris image shall be accompanied by an iris image header, that contains an image sequence number plus information about quality and rotation for that image. Each image shall be padded with extra bits, if necessary, to end on an integral byte boundary. All header data shall be stored in network byte (big-endian) order. Where bit-level data definitions are specified, bit 1 shall be interpreted as the least significant bit (LSB). Signed fields shall use 2's complement encoding. A single record header structure shall be used for both rectilinear and polar formats. An iris image data record shall contain either rectilinear format images or polar format images; the two formats shall not be intermixed within a data record. The CBEFF Format Type value in the CBEFF header shall indicate the format used, with the value 0x0601 used to indicate rectilinear coordinates and 0x0602 to indicate polar coordinates as specified in 5.1.
Table 1 — Iris image data record
Bytes Type Content Description
1 - 63 Iris Record Header Information pertaining to capture device,
number of features (eyes) included in the
record, and total record size in bytes
64 - 66 Iris Feature Header Header for feature indicating right or left eye,
if known, and number of images
67 - 81 Iris Image Header Header for first image containing image
number for first eye, quality, rotation angle
and uncertainty, and image length
82 – (image
length-1)
Unsigned char Image First image, this eye
Iris Image Header Image header, second image, first eye
Unsigned char Image Second image, first eye
• •
• • • •
Iris Image Header Image header, last image, first eye Unsigned char Image Last image, first eye
Iris Feature Header Feature header indicating right or left eye
and number of images (used only if eye type
is known)
Iris Image Header Image header, first image, second eye
Unsigned char Image First image, second eye
Iris Image Header Image header, second image, second eye Unsigned char Image Second image, second eye
• •
• • • •
Iris Image Header Image header, last image, second eye
Unsigned char Image Last image, second eye
5.5 Iris header structures
5.5.1 Iris Record Header structure - The iris record header shall contain data values that identify the record as containing iris image data, specify properties of the image capture device and the image format, and designate
a unique identifier for the data capture event. The record header format is shown in Table 2.
5.5.2 Iris Feature Header structure – The iris feature header shall contain data values that indicate which eye was presented, if known, and the number of images captured for that eye. The iris feature header format is shown in Table 3.
5.5.3 Iris Image Header structure – The iris image header shall indicate an image sample number, quality value, eye rotation information (if available), and image data length. It will be followed by the image data. The iris image header format is shown in Table 4.
5.5.4 CBEFF Product ID - The ‘owner’ of the product or device used to generate the image data (or transforming application) shall be uniquely identified by the two most significant bytes of a CBEFF Product ID (Owner). The product software version or device version shall be uniquely identified as the least significant two bytes of the CBEFF Product ID (Type). The type value of the CBEFF Product ID is established by the vendor. The value can be obtained from the IBIA. The CBEFF Product ID is specified in the augmented version of CBEFF (NISTIR 6529-A). CBEFF Product ID codes will be assigned by the International Biometric Industry Association (). Refer to the CBEFF standard for the definition of CBEFF Format Owner, Format Type, and Product ID, and information on how to obtain a CBEFF Product ID from IBIA.
5.5.5 Capture Device ID – The iris record header shall include a two byte identification code for the image capture device that generated the image data. This code shall be vendor-specified.
5.5.6 Iris Image Quality – The quality of the rectilinear iris image or the quality of the original image used to produce the polar image shall be represented in a five-byte set of fields. The quality score shall occupy the first byte. The value of the quality score shall range from 0 to 100 for samples where the sample quality has been assessed. A value of 254 shall indicate that no attempt was made to assess the sample quality. A value of 255 shall indicate that the sample quality assessment was attempted but was unsuccessful (e.g. due to iris segmentation failure). Values of biometric sample quality must be interpreted in view of the method used to assess the quality. Therefore, the quality assessment algorithm’s supplier (owner) and specific identity (ID) are contained in the next four bytes. The quality algorithm ID is the 16-bit format owner value assigned by the International Biometric Industry Association. The next two bytes contain a specific identifier for the quality assessment algorithm. This value is assigned by the organization, and may be registered with the IBIA.。