ChapterPDF Available

Signature/Sign Time Series Data: Standardization

Authors:

Figures

Content may be subject to copyright.
Encyclopedia of Biometrics
DOI 10.1007/978-3-642-27733-7_9125-2
© Springer Science+Business Media New York 2014
Signature/Sign Time Series Data: Standardization
Olaf Hennigera,RichardGuest
b,Oscar Miguel-Hurtadoband Christiane Kaplanc
aFraunhofer Institute for Computer Graphics Research IGD, Darmstadt, Germany
bSchool of Engineering and Digital Arts, University of Kent, Canterbury, Kent, UK
cSOFTPRO GmbH, Böblingen, Germany
Synonyms
Online signature data format;Signature/sign behavioral data interchange format
Definition
The International Standard ISO/IEC 19794-7 specifies data interchange formats for online
signature data in the form of multidimensional time series. These formats support interoperability
and data interchange among the subsystems of open online signature recognition systems.
Overview
Format Types
This entry elaborates on the latest edition of ISO/IEC 19794-7 developed by Joint Technical
Committee ISO/IEC JTC 1 Subcommittee SC 37, the committee responsible for the international
standardization of generic biometric technologies. ISO/IEC 19794-7 [1,2] specifies the following
data formats for online signature time series data representing signature dynamics (The text taken
from ISO/IEC 19794-7 is reproduced with the permission of the International Organization for
Standardization, ISO. This standard can be obtained from any ISO member body and from the
web site of the ISO Central Secretariat at the following address www.iso.org. Copyright remains
with ISO.):
Full format
Compression format
Compact format
XML-based format
The full format may be applied in a wide range of application areas where handwritten online
signatures or signs are involved. The compression format is capable of holding the same informa-
tion as the full format, but in losslessly compressed form. The compact format is designed for use
with smart cards and other tokens. It does not require resources for compression/decompression,
E-mail: olaf.henniger@igd.fraunhofer.de
Page 1 of 9
Encyclopedia of Biometrics
DOI 10.1007/978-3-642-27733-7_9125-2
© Springer Science+Business Media New York 2014
but reduces the record size by conveying information quantized at a lower resolution. While the
full, compression, and compact format all are binary formats, the XML-based format is a textual
format. The XML-based format is intended for data interchange over web interfaces.
The time series data formats can be used for both unprocessed and processed online signature
samples. Unprocessed time series data come directly from capture devices such as touch screens,
digitizing tablets, or special pens. The time series data may be processed in order to render
signatures comparable with each other. Possible processing steps include low-pass filtering,
parallel translation and rotation of signatures, and normalization and scaling.
Which of the format types and which options are to be applied in a particular application should
be defined in application-specific requirement specifications or application profiles.
Channels
Online signatures may contain different data items recorded in form of time series. Each such data
item is referred to as a channel. The following channels may be utilized:
1. xcoordinate (X)
2. ycoordinate (Y)
3. zcoordinate (Z)
4. Velocity in xdirection (VX)
5. Velocity in ydirection (VY)
6. Acceleration in xdirection (AX)
7. Acceleration in ydirection (AY)
8. Elapsed time since the first sample point (T)
9. Elapsed time since the previous sample point (DT)
10. Magnitude of the pen tip force (F)
11. Pen tip switch state (S)
12. Pen tilt along the xaxis (TX)
13. Pen tilt along the yaxis (TY)
14. Pen azimuth (A)
15. Pen elevation (E)
16. Pen rotation (R)
If present, the channels are included in the order given above. Either the T channel or the DT
channel must be present, or uniform sampling (constant time difference between adjacent sample
points) must be indicated. The inclusion of at least one of the other channels is mandatory. The
sample frequency should be at least 50 samples per second.
Coordinate System
The coordinate system for expressing the pen position is a three-dimensional Cartesian coordinate
system. The xaxis is the horizontal axis of the writing plane, with xcoordinates increasing to the
right. The yaxis is the vertical axis of the writing plane, with ycoordinates increasing upwards.
The origin of xand ycoordinates depends on the used technology. It may be, for instance, at the
lower left corner of the writing plane, at the pen position at the first pen-down event, or in the
Page 2 of 9
Encyclopedia of Biometrics
DOI 10.1007/978-3-642-27733-7_9125-2
© Springer Science+Business Media New York 2014
center of a signature. The zaxis is the axis perpendicular to the writing plane, with zcoordinates
increasing upwards out of the writing plane starting from 0. The zcoordinate may be used for
reporting the height of the pen above the writing plane.
Full Format
Structure of a Data Record
An online signature time series data record in the full format as defined in [1] consists of the
following data elements in the given order:
General header
Record body
General Header
The general header consists of the following data elements in the given order:
Format identifier character string (“SDI”)
Version number character string
Length of the data record in bytes
Number of subsequent signature/sign representations
Certification flag
The version number indicates which version of ISO/IEC 19794 ([1]or[3]) applies. The
certification flag must be set to 0. It has been added for upward compatibility with later versions
of the format that may provide space for certification information.
Record Body
Structure of the Record Body
The record body consists of a sequence of at least one signature/sign representation. A repre-
sentation refers to a sub-record that contains time series data of a single online signature. Each
signature/sign representation consists of the following data elements in the given order:
Representation header
Representation body
Representation Header
Structure of the Representation Header The representation header contains representation-
specific descriptive information. It consists of the following data elements in the given order:
Page 3 of 9
Encyclopedia of Biometrics
DOI 10.1007/978-3-642-27733-7_9125-2
© Springer Science+Business Media New York 2014
Length of the representation in bytes
Capture date and time in Coordinated Universal Time (UTC),
Capture device technology identifier
Capture device vendor identifier
Capture device type identifier
Quality record
Sequence of channel descriptions
Number of sample points
Capture Device Technology, Vendor, and Type Identifiers If not set to 0, these identifiers
indicate:
The class of capture device technology (electromagnetic, semiconductor, special pen with
acceleration sensors, or special pen with optical sensors) used to acquire the online signature
The biometric organization that owns the product that created the data record
The product type that created the data record
The capture device vendor identifier must be registered by the Biometric Registration Authority
in order to provide for unambiguous identification.
Quality Record The quality record consists of a length field representing the number of
subsequent quality blocks followed by zero or more quality blocks. Each quality block consists
of the following data elements in the given order:
Quality score
Quality algorithm vendor identifier
Quality algorithm identifier
The quality score expresses the predicted comparison performance of the signature/sign represen-
tation as a value between 0 and 100, with higher values indicating better quality. If not set to 0,
the quality algorithm vendor identifier and the quality algorithm identifier identify the provider
of the algorithm and the algorithm that created the quality score, respectively. In order to provide
for unambiguous identification, the quality algorithm vendor identifier must be registered by the
Biometric Registration Authority.
Channel Descriptions The channel descriptions field begins with a channel inclusion field
indicating the presence or absence of channels. Each bit corresponds to a channel. While a bit
value of 1 encodes the presence of the corresponding channel, a bit value of 0 encodes the absence
of the corresponding channel.
The channel inclusion field is followed by a sequence of channel descriptions for the channels
indicated as present in the channel inclusion field. The channel descriptions are mandatory for all
channels present in the data record. Each channel description begins with a preamble. The channel
description preamble encodes the presence of the channel attributes:
Scaling value
Minimum possible channel value
Page 4 of 9
Encyclopedia of Biometrics
DOI 10.1007/978-3-642-27733-7_9125-2
© Springer Science+Business Media New York 2014
Maximum possible channel value
Average of the channel values
Empirical standard deviation of the channel values
The channel description preamble of the DT channel may indicate that the value of this channel is
constant, i.e., that the sampling is equidistant in time. In this case, the DT channel is absent in the
representation body even though the channel inclusion field indicates its presence. If the channel
description contains a scaling value, then the constant time difference between adjacent sample
points is 1 divided by the scaling value.
Furthermore, the channel description preamble may indicate that the linear component of the
linear regression model over time has been removed from the channel. For instance, the linear
component of the X-on-T regression line (due to writing along a horizontal line in most writing
systems) may be removed during preprocessing of online signatures in order to map the X values
to a smaller range.
The channel description preamble is followed by those channel attributes indicated as present.
Each channel attribute is encoded in 2 bytes. Scaling values are represented as pairs of base two
exponent and mantissa. The exponent is encoded in 5 bits. Thus, signed integer values in the range
from 16 to 15 are possible for the exponent. The mantissa lies in the range from 1 to 2 with up
to 11 positions after the binary point in binary notation. The channel values as well as the other
channel attributes are to be divided by the corresponding scaling value in order to restore their
actual values in the corresponding unit of measurement. By choosing appropriate scaling values,
different degrees of accuracy can be expressed. If the scaling value is absent, the calibration of the
corresponding channel is unknown.
If present, the other channel attributes provide information as to whether, and to which range,
the online signature has been transformed during preprocessing for facilitating comparison. For
instance, the averages of the X and Y channels may provide information as to whether the online
signature has been shifted and rotated.
Representation Body
A representation body consists of the following data elements in the given order:
Sequence of sample point fields, one for each sample point
Extended data length in bytes possibly followed by extended data
Each sample point field consists of a sequence of channel values as indicated by the channel
inclusion field in the representation header. For all channels except for the S channel (recording
whether the pen tip touches the writing plane or not), channel values are encoded as 2-byte integers.
Signed values are biased by 32,768, i.e., the actual value is 32,768 less than the stored value. For
the S channel, only the values 0 and 1 are allowed and the channel values are encoded in 1 byte.
The optional extended data may be of any format. If extended data of unknown format is present,
it may be ignored.
Page 5 of 9
Encyclopedia of Biometrics
DOI 10.1007/978-3-642-27733-7_9125-2
© Springer Science+Business Media New York 2014
Compression Format
Structure of a Data Record
Like in the full format, a data record in the compression format consists of a general header and
a record body. The general header is as described for the full format. Just the format identifier is
different (“SCD”).
Record Body
Structure of the Record Body
Like in the full format, the record body consists of a sequence of at least one signature/sign
representation, and each signature/sign representation consists of a representation header and a
representation body.
Representation Header
The representation header is as described for the full format, with the following additional data
elements appended at the end:
Compression algorithm identifier
Length of the compressed data in bytes
The compression algorithm identifier indicates the lossless compression algorithm used (Bzip2,
LZW, GZip, Deflate, PPMd, LZMA, or Zip are the designated options). All these compression
algorithms reduce the record size by about half, similar to what the compact format achieves, but
without loss of information [4].
Representation Body
A representation body in compression format consists of the following data elements in the given
order:
Block of compressed data
Extended data length in bytes possibly followed by extended data
Unlike the data in the record body of the full format, which is arranged as a sequence of sample
point fields, one for each sample point, the data to be compressed is arranged as a sequence
of difference channels, one for each channel indicated as present, because this yields a better
compression rate. Every difference channel starts with the initial value c1of the channel. This
is followed by a sequence of differences diDciC1ci.1 iN1/ between values at
consecutive sample points (Nis the number of sample points). Each difference is encoded as a 2-
byte integer. The sequence of difference channels is compressed using the compression algorithm
indicated in the representation header.
Page 6 of 9
Encyclopedia of Biometrics
DOI 10.1007/978-3-642-27733-7_9125-2
© Springer Science+Business Media New York 2014
Compact Format
Overview
The compact format of online signature time series data may be applied for both off-card and on-
card comparison. The compact format of the second edition [1] is the same as the one defined in
the first edition of ISO/IEC 19794-7 [3].
The compact format is more compact than the full format: An online signature time series data
block in compact format does not contain a header. Information about the structure and contents of
the data record that otherwise would be given in the header is contained in a separate data object,
referred to as comparison algorithm parameters (see below). An online signature time series data
record in compact format encodes each channel value within 1 byte only and contains only one
signature/sign representation. While the conversion of the channel values from 2 bytes to 1 byte
reduces the record size by half, the higher quantization error leads to only a marginal increase of
the error rates of signature comparison algorithms based on Dynamic Time Warping [4].
Comparison Algorithm Parameters Template
The comparison algorithm parameters data object may be embedded in a Biometric Header
Template (BHT), which itself is embedded in a Biometric Information Template (BIT) as defined
in the TLV- (tag-length-value) encoded CBEFF patron format for use with smart cards or other
tokens [5]. The comparison algorithm parameters data object may include:
The maximum number of sample points that the comparison algorithm is able to process.
A sequence of channel descriptions, as described above for the full format. In contrast to the full
format, the channel attributes with the exception of the scaling values are encoded in 1 byte.
For illustration, Fig. 1shows the annotated hex dump of the comparison algorithm parameters data
object associated with the data block depicted in Fig. 2
Tag of comparison algorithm parameters data object
Length of comparison algorithm parameters data object: 9 bytes
Tag of channel descriptions
Length of channel decriptions: 7 bytes
X, Y, and DT channels included
X channel: No information given
Y channel: No information given
DT scaling value included, uniform sampling
DT scaling value: 50 samples/s (base 2 exponent: 5, mantissa: 1,1001b)
B1
09
86
07
C0 80
00
00
84
AC 80
Fig. 1 Annotated hex dump of a comparison algorithm parameters data object
Beside the comparison algorithm parameters data object, the BHT should also include a format
owner identifier and a format type identifier for the associated biometric data. The format type
identifier for the online signature time series compact format is 15 (000fHex). The format owner
identifier is 257 (0101Hex), representing the standardization committee ISO/IEC JTC 1/SC 37.
These identifiers are registered by the Biometric Registration Authority.
Page 7 of 9
Encyclopedia of Biometrics
DOI 10.1007/978-3-642-27733-7_9125-2
© Springer Science+Business Media New York 2014
Embedment in a CBEFF Data Structure
An online signature time series data block in compact format must be embedded in a CBEFF data
structure in the TLV-encoded patron format for use with smart cards or other tokens defined in
[5]. Its tag is defined in [5]. Its length is encoded following the Distinguished Encoding Rules of
ASN.1 [6]. Lastly, its value is the record body described below.
Record Body
Like in the full format, the record body consists of a sequence of sample point fields, one for each
sample point. Each sample point field consists of a sequence of channel values as indicated by the
channel inclusion field in the comparison algorithm parameters data object. In contrast to the full
format, however, all channel values are encoded in 1 byte only and signed values are biased by
128.
For illustration, Fig. 2shows the annotated hex dump of the initial part of an online signature
time series data block in compact format.
Tag of biometric data block without extended data
1 length byte follows
Length of biometric data block: 178 bytes
Sample point 1: X = 112, Y = 71
Sample point 2: X = 106, Y = 63
and so on for the remaining 178/2 – 2 =87 sample points
5F 2E
81
B2
F0 C7
EA BF
...
Fig. 2 Annotated hex dump of the initial part of an online signature time series data block in compact format
XML Encoding
The XML schema in [2] describes the structure of XML documents containing online signature
time series data. All XML data elements map to data elements in the full format of [1].
However, some data elements from the full format (such as record length in bytes and number of
representations) that help in parsing binary data records, but are dispensable in XML documents,
do not have a counterpart in the XML schema.
The Ink Markup Language developed by the World Wide Web Consortium (W3C) [7] provides
another XML-based format for online signature time series data.
Related Entries
Biometric Data Interchange Formats, Standardization
Signature Comparison
Signature Features
Signature Recognition, Overview
Page 8 of 9
Encyclopedia of Biometrics
DOI 10.1007/978-3-642-27733-7_9125-2
© Springer Science+Business Media New York 2014
References
1. Information technology – Biometric data interchange formats – Part 7: Signature/sign time series
data. International Standard ISO/IEC 19794-7, second edition, 2014. Available at http://www.
iso.org/iso/home/store.htm
2. Information technology – Biometric data interchange formats – Part 7: Signature/sign time series
data – Amendment 1: XML Encoding. Draft Amendment to [1] (2014) Under development
3. Information technology – Biometric data interchange formats – Part 7: Signature/sign time series
data. International Standard ISO/IEC 19794-7, first edition, 2007. Available at http://www.iso.
org/iso/home/store.htm
4. O. Miguel-Hurtado, Online signature verification algorithms and development of signature
international standards. PhD thesis, Universidad Carlos III de Madrid, 2011
5. Information technology – Common Biometric Exchange Formats Framework – Part 3: Patron
format specifications. International Standard ISO/IEC 19785-3. Available at http://www.iso.
org/iso/home/store.htm
6. Information technology – ASN.1 encoding rules – Part 1: Specification of Basic Encoding
Rules (BER), Canonical Encoding Rules (CER) and Distinguished Encoding Rules (DER).
International Standard ISO/IEC 8825-1. Available at http://www.iso.org/iso/home/store.htm
7. Ink Markup Language (InkML). W3C Recommendation, 2011. Available at http://www.w3.
org/TR/InkML
Page 9 of 9
ResearchGate has not been able to resolve any citations for this publication.
Thesis
Full-text available
The science of biometrics is based on discovering the identities of human beings by investigating their physical and behavioural traits. Of the many different biometric traits, i.e. fingerprint, iris, vascular, etc... the handwritten signature is still one of the most accepted techniques. Advancing progress in identification applications has led to widespread demand for new generation ID documents, such as electronic passports and citizen cards, which contain additional biometric information required for more accurate user recognition. This can be achieved by embedding dynamic signature features within the documentation. However, this would result in two significant drawbacks that must be addressed, these are: Memory Capacity and Computational Load. These problems and the increasing demand for standardized biometric verifications systems have motivated the research work performed in this Thesis. In order to achieve this, an attempt to reduce the information involved in verification processes is performed using feature selection criteria of the signature biometric data. Such reduced information content not only satisfies the memory capacity restrictions but also provides much more efficient use of the verification algorithms. In particular, two novel methods in the signature context, based on Principal Component Analysis and Hellinger Distance, are proposed here. The performance of the optimized features set obtained has been analyzed using two different verification algorithms. By reducing the sample size it has been observed that the error rates are maintained sufficiently low and the results obtained are in agreement with the current state of the art for signature techniques. It will be shown that in some cases that feature selection does not provide an adequate reduction solution, where a different strategy has been analyzed to achieve the aforementioned problems. A direct consequence of the widespread nature of biometric verification has led to demands for standardized protocols to improve interoperability. The work presented throughout this Thesis has considered current ISO/IEC signature standard data formats. It has been observed that the current compact data formats, 19794-7 Compact Format and 19794- 11, do not meet the requirements of modern data formats. In particular, 19794-7 Compact Format, although having good compression ratios, has been found to imply an inadmissible loss in information. This problem has been solved by defining a new near-lossless compression data format based on lossless compression algorithms, and proposing different enhanced strategies to store signature data. This new data format achieves the same compression ratio, but without losing any relevant information. In addition, the problems found in the 19794-11CD2 regarding the lack of compression and information loss have been addressed. A new data format structure has been proposed, where the lack of compression is solved by reducing the data stored, avoiding duplicated data and providing a new singular point definition. This new structure has provided improved compression ratios, and, at the same time, carries more information. The two new data format definitions were presented to the ISO/IEC SC37 WG3 experts and accepted as the new third subformat “Compression Format” within the 19794-7 and the new committee draft for the 197974-11 CD3.
Information technology – Biometric data interchange formats – Part 7: Signature/sign time series data – Amendment 1: XML Encoding. Draft Amendment to [1] (2014) Under development 3. Information technology – Biometric data interchange formats – Part 7: Signature/sign time series data
Information technology – Biometric data interchange formats – Part 7: Signature/sign time series data – Amendment 1: XML Encoding. Draft Amendment to [1] (2014) Under development 3. Information technology – Biometric data interchange formats – Part 7: Signature/sign time series data. International Standard ISO/IEC 19794-7, first edition, 2007. Available at http://www.iso. org/iso/home/store.htm