ISO/IEC TR 15938-8:2002/Amd 5:2010
(Amendment)Information technology — Multimedia content description interface — Part 8: Extraction and use of MPEG-7 descriptions — Amendment 5: Extraction and matching of image signature tools
Information technology — Multimedia content description interface — Part 8: Extraction and use of MPEG-7 descriptions — Amendment 5: Extraction and matching of image signature tools
Technologies de l'information — Interface de description du contenu multimédia — Partie 8: Extraction et utilisation des descriptions MPEG-7 — Amendement 5: Extraction et correspondance des outils de signature d'image
General Information
Relations
Standards Content (Sample)
TECHNICAL ISO/IEC
REPORT TR
15938-8
First edition
2002-12-15
AMENDMENT 5
2010-03-15
Information technology — Multimedia
content description interface —
Part 8:
Extraction and use of MPEG-7
descriptions
AMENDMENT 5: Extraction and matching of
image signature tools
Technologies de l'information — Interface de description du contenu
multimédia —
Partie 8: Extraction et utilisation des descriptions MPEG-7
AMENDEMENT 5: Extraction et correspondance des outils de signature
d'image
Reference number
ISO/IEC TR 15938-8:2002/Amd.5:2010(E)
©
ISO/IEC 2010
---------------------- Page: 1 ----------------------
ISO/IEC TR 15938-8:2002/Amd.5:2010(E)
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.
COPYRIGHT PROTECTED DOCUMENT
© 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
---------------------- Page: 2 ----------------------
ISO/IEC TR 15938-8:2002/Amd.5:2010(E)
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.
International Standards are drafted in accordance with the rules given in the ISO/IEC Directives, Part 2.
The main task of the joint technical committee is to prepare International Standards. Draft International
Standards adopted by the joint technical committee are circulated to national bodies for voting. Publication as
an International Standard requires approval by at least 75 % of the national 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 and IEC shall not be held responsible for identifying any or all such patent rights.
Amendment 5 to ISO/IEC TR 15938-8:2002 was prepared by Technical Committee ISO/IEC JTC 1,
Information technology, Subcommittee SC 29, Coding of audio, picture, multimedia and hypermedia
information.
© ISO/IEC 2010 – All rights reserved iii
---------------------- Page: 3 ----------------------
ISO/IEC TR 15938-8:2002/Amd.5:2010(E)
Information technology — Multimedia content description
interface —
Part 8:
Extraction and use of MPEG-7 descriptions
AMENDMENT 5: Extraction and matching of image signature tools
Add after 4.8:
4.9 Visual Signatures
4.9.1 Image Signature
The visual content descriptors in ISO/IEC 15938-3 clauses 6 to 9 are very useful when trying to find images
with similar content. However, such descriptors are intended to be general and are found to be unsuitable for
the task of finding duplicate images. The image signature descriptor is robust (unchanging) across a wide
range of common editing operations, but is sufficiently different for every item of "original" content to identify it
uniquely and reliably – just like human fingerprints.
There are three components within the image signature, the first two are global signatures, representing the
complete image and the third is composed of a set of local signatures, each representing part of the image.
4.9.1.1 Feature Extraction
11.3.5 and 11.3.6 of ISO/IEC 15938-3:2002/Amd.3:2009 describe the extraction of the image signature.
4.9.1.2 Similarity Matching
4.9.1.2.1 Global Signatures
To perform matching between two global signatures B and B both of length N, the Hamming distance
1 2
should be taken:
H(,BB)=⊗B B ,
12 ∑ 1 2
where ⊗ denotes the exclusive OR (XOR) operator.
This can be normalised to the range 0-1 using the normalised Hamming distance:
1
ˆ
H(,BB)=⊗B B .
12 ∑ 1 2
N
Evaluation on test data has shown that for independent images a Hamming distance H(,BB) of 147
12
corresponds to a false alarm rate of less than 0.05 parts per million (ppm). A distance of 169 corresponds to a
false alarm rate of 1 ppm.
© ISO/IEC 2010 – All rights reserved 1
---------------------- Page: 4 ----------------------
ISO/IEC TR 15938-8:2002/Amd.5:2010(E)
4.9.1.2.2 Local Signatures
1. Form hypotheses of 3 features
Use 32 out of 80 features
2. Apply Geometric Constraints
Feature direction & ratio of line length
3. Form hypotheses of 4 features
Use all 80 features
4. Apply Geometric Constraint
Ratio of areas
Figure AMD5.1 — Feature matching is carried out in a four stage process which combines hypothesis
forming (using the local signature) and geometric constraints
Matching images can be performed by comparing the local signatures which form a part of the image
signature. For efficiency four stages can be used, as shown in Figure AMD5.1, to match local signatures from
a pair of images. Hypotheses can be formed in stages one and three. A series of geometric test are performed
in stages two and four, these tests should be passed in order for a hypothesis to progress to the next stage.
The stages become increasingly computationally complex so each stage aims to minimise the number of
hypotheses that are accepted for subsequent processing.
The first stage involves forming hypotheses for potential matching feature points by comparing the first 32
feature points from a first image with the first 32 feature points from a second image. To perform matching
between feature points the local signatures (B and B ) for the feature points can be compared using the
1 2
Hamming distance:
H (B ,B ) = B ⊗B .
1 2 ∑ 1 2
A list of feature point pairs with a Hamming distance less than a predefined threshold T is created. A set of
A
hypotheses (candidate pairs of potentially matching features) is generated by taking all combinations of three
pairs of matching feature points from the list. A hypothesis is made up of a set of three pairs of feature points
( (a ,b ),(ab, ),(ab, )), where aa,,a are feature points from a first image and bb,,b are feature
11 2 2 3 3 12 3 12 3
points from a second image. A set of three pairs of feature points is declared a match if the cumulative
Hamming distance between the local signatures, of corresponding feature points, is below a constant
threshold T . The set of hypotheses is ordered by their cumulative distance, with the lowest distance (i.e.
B
highest likelihood of being a match) first. In order to minimise complexity the number of hypotheses may be
limited to the hypotheses corresponding to the lowest distance.
A second stage applies geometric constraints to each hypothesis generated from stage one.
...
Questions, Comments and Discussion
Ask us and Technical Secretary will try to provide an answer. You can facilitate discussion about the standard in here.