Download Full Multimedia Security Watermarking Steganography and Forensics 1st Edition Frank Y. Shih (Editor) PDF All Chapters

Download the full version of the ebook at
https://ebookultra.com
Multimedia Security Watermarking

Steganography and Forensics 1st Edition Frank
Y. Shih (Editor)
https://ebookultra.com/download/multimedia-
security-watermarking-steganography-and-
forensics-1st-edition-frank-y-shih-editor/
Explore and download more ebook at https://ebookultra.com

Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.
Disappearing cryptography Information hiding Steganography

watermarking 3ed. Edition Wayner P.
https://ebookultra.com/download/disappearing-cryptography-information-
hiding-steganography-watermarking-3ed-edition-wayner-p/
ebookultra.com
Multimedia Security Handbook 1st Edition Borko Furht
https://ebookultra.com/download/multimedia-security-handbook-1st-
edition-borko-furht/
ebookultra.com
Android Forensics Investigation Analysis and Mobile

Security for Google Android 1st Edition Andrew Hoog
https://ebookultra.com/download/android-forensics-investigation-
analysis-and-mobile-security-for-google-android-1st-edition-andrew-
hoog/
ebookultra.com
Cyber Security Understanding Cyber Crimes Computer

Forensics And Legal Perspectives 1st Edition Wiley India
https://ebookultra.com/download/cyber-security-understanding-cyber-
crimes-computer-forensics-and-legal-perspectives-1st-edition-wiley-
india/
ebookultra.com
Network Forensics 1st Edition Messier
https://ebookultra.com/download/network-forensics-1st-edition-messier/
ebookultra.com
The Facebook era 4. print Edition Shih
https://ebookultra.com/download/the-facebook-era-4-print-edition-shih/
ebookultra.com
Exploratory Image Databases AP Communications Networking

and Multimedia Series Communications Networking and
Multimedia 1st Edition Simone Santini
https://ebookultra.com/download/exploratory-image-databases-ap-
communications-networking-and-multimedia-series-communications-
networking-and-multimedia-1st-edition-simone-santini/
ebookultra.com
Learn Chinese from modern writers C. W. Shih
https://ebookultra.com/download/learn-chinese-from-modern-writers-c-w-
shih/
ebookultra.com
Microbial Forensics 1st Edition Roger G. Breeze
https://ebookultra.com/download/microbial-forensics-1st-edition-roger-
g-breeze/
ebookultra.com
Multimedia Security Watermarking Steganography and
Forensics 1st Edition Frank Y. Shih (Editor) Digital
Instant Download
Author(s): Frank Y. Shih (Editor)
ISBN(s): 9781439873328, 1351824120
Edition: 1
File Details: PDF, 11.51 MB
Year: 2013
Language: english
Electrical Engineering/Computer Science shih
MultiMedia Security:
Watermarking, Steganography, Watermarking,
and Forensics
Steganography, and Forensics
Multimedia Security: Watermarking, Steganography, and Forensics outlines
essential principles, technical information, and expert insights on multimedia security
technology used to prove that content is authentic and has not been altered. Illustrat-
ing the need for improved content security as the Internet and digital multimedia appli-
cations rapidly evolve, this book presents a wealth of everyday protection application
examples in fields including multimedia mining and classification, digital watermark-
ing, steganography, and digital forensics.
Giving readers an in-depth overview of different aspects of information

security mechanisms and methods, this resource also serves as an instructional tool
on how to use the fundamental theoretical framework required for the development
of extensive advanced techniques. The presentation of several robust algorithms
illustrates this framework, helping readers to quickly master and apply fundamental
principles.
Presented case studies cover:
• The execution (and feasibility) of techniques used to discover hidden

knowledge by applying multimedia duplicate mining methods to large
and Forensics
Watermarking, Steganography,
multimedia content
• Different types of image steganographic schemes based on vector
quantization
• Techniques used to detect changes in human motion behavior and to
classify different types of small-group motion behavior
Useful for students, researchers, and professionals, this book consists of a variety of
technical tutorials that offer an abundance of graphs and examples to powerfully con- Edited By
vey the principles of multimedia security and steganography. Imparting the extensive
experience of the contributors, this approach simplifies problems, helping readers
more easily understand even the most complicated theories. It also enables them to
uncover novel concepts involved in the implementation of algorithms, which can lead
to the discovery of new problems and new means of solving them.
K13398
6000 Broken Sound Parkway, NW

Suite 300, Boca Raton, FL 33487
711 Third Avenue
New York, NY 10017
2 Park Square, Milton Park
Abingdon, Oxon OX14 4RN, UK
www.crcpress.com
fiJHRUijJM SI:C:lJI=tiTY
'"'ATI:I:ttv1AI:ti<INC;.
STI:C;ANC)C;I:tAili-IY. ANil i=C)I:ti:NSIC:S
~CRC Press
~ Taylor&FrancisGroup
Boca Raton London New York
CRC Press is an imprint of the

Taylor & Francis Group, an informa business
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2013 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works

Version Date: 20120531
International Standard Book Number-13: 978-1-4398-7332-8 (eBook - PDF)
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been
made to publish reliable data and information, but the author and publisher cannot assume responsibility for the
validity of all materials or the consequences of their use. The authors and publishers have attempted to trace the
copyright holders of all material reproduced in this publication and apologize to copyright holders if permission to
publish in this form has not been obtained. If any copyright material has not been acknowledged please write and let
us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted,
or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, includ-
ing photocopying, microfilming, and recording, or in any information storage or retrieval system, without written
permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.com
(http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers,
MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety
of users. For organizations that have been granted a photocopy license by the CCC, a separate system of payment
has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for
identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at

http://www.crcpress.com
To my loving wife and children, to my parents who encouraged me
through the years, and to those who helped in the process of writing
this book
This page intentionally left blank
Contents
Preface.......................................................................................................................ix
Acknowledgments................................................................................................... xiii
Editor........................................................................................................................ xv
Contributors............................................................................................................xvii
Part I Multimedia Mining and Classification
Chapter 1 Multimedia Duplicate Mining toward Knowledge Discovery..............3

Xiaomeng Wu, Sebastien Poullot, and Shin’ichi Satoh
Chapter 2 Discriminative Learning-Assisted Video Semantic Concept

Classification....................................................................................... 31
Qiusha Zhu, Mei-Ling Shyu, and Shu-Ching Chen
Chapter 3 Improved Feature Vocabulary-Based Method for Image

Categorization..................................................................................... 51
Frank Y. Shih and Alexander Sheppard
Part II Watermarking
Chapter 4 Automatic Detection and Removal of Visible Image Watermarks..... 67
Hong-Ren Su, Ya-Yun Cheng, and Shang-Hong Lai
Chapter 5 Digital Watermarking Based on Chaotic Map and Reference

Register............................................................................................ 85
Yi-Ta Wu and Frank Y. Shih
Chapter 6 Pseudo-Random Pixel Rearrangement Algorithm Based on

Gaussian Integers for Image Watermarking.......................................99
Aleksey Koval, Frank Y. Shih, and Boris S. Verkhovsky
Chapter 7 Reversible Data-Hiding Techniques for Digital Images................... 115

Zhi-Hui Wang, Ming-Ting Sun, and Chin-Chen Chang
v
vi Contents
Chapter 8 Watermarking Based on Local Binary Pattern Operators................ 141

Wenyin Zhang and Frank Y. Shih
Chapter 9 Authentication of JPEG Images Based on Genetic Algorithms....... 165

Venkata Gopal Edupuganti and Frank Y. Shih
Chapter 10 An Efficient Block-Based Fragile Watermarking System

for Tamper Localization and Recovery............................................. 181
Venkata Gopal Edupuganti, Frank Y. Shih, and I-Cheng Chang
Part III Steganography
Chapter 11 Survey of Image Steganography and Steganalysis............................ 201

Mayra Bachrach and Frank Y. Shih
Chapter 12 Digital Steganographic Schemes Based on Image Vector

Quantization.....................................................................................215
Shinfeng D. Lin and Shih-Chieh Shie
Chapter 13 Differential Evolution-Based Algorithm for Breaking

the Visual Steganalytic System......................................................... 245
Frank Y. Shih and Venkata Gopal Edupuganti
Chapter 14 Genetic Algorithm-Based Methodology for Breaking the

Steganalytic Systems......................................................................... 257
Yi-Ta Wu and Frank Y. Shih
Part IV Forensics
Chapter 15 Image Inpainting Using an Enhanced Exemplar-Based

Algorithm.......................................................................................... 275
I-Cheng Chang and Chia-We Hsu
Chapter 16 A Comparison Study on Copy–Cover Image Forgery Detection..... 297

Frank Y. Shih and Yuan Yuan
Contents vii
Chapter 17 Chaos-Based Hash Function with Both Modification

Detection and Localization Capabilities........................................... 313
Di Xiao, Frank Y. Shih, and Xiaofeng Liao
Chapter 18 Video Forensics................................................................................. 327

Hsiao-Rong Tyan and Hong-Yuan Mark Liao
Chapter 19 Using the Self-Synchronizing Method to Improve Security

of the Multi-Chaotic Systems-Based Image Encryption................... 343
Di Xiao and Frank Y. Shih
Chapter 20 Behavior Modeling of Human Objects in Multimedia Content........ 357

Yafeng Yin and Hong Man
Index �� 385

Preface
Digital watermarking, steganography, and forensics are important topics because
digital multimedia are widely used and the Internet is rapidly growing. This book
intends to provide a comprehensive overview of the different aspects of mechanisms
and techniques for information security. It has been written for students, researchers,
and professionals who take related courses, want to improve their knowledge, and
want to gain experience pertaining to the role of digital watermarking, steganogra-
phy, and forensics.
Digital watermarking technology can be used to guarantee authenticity and can
be applied as proof that the content has not been altered since insertion.
Steganographic messages are often first encrypted by some traditional means,
and then a covert text is modified in some way to contain the encrypted
message.
Digital forensics utilizes computational techniques to process multimedia content
and then applies the results to help crime investigations. The need for information
security exists everywhere every day.
This book aims to provide students, researchers, and professionals with techni-
cal information regarding digital watermarking, steganography, and forensics, as
well as instruct them on the fundamental theoretical framework in developing the
extensive advanced techniques. By comprehensively considering the essential prin-
ciples of digital watermarking, steganography, and forensics systems, one can not
only obtain novel ideas in implementing the advanced algorithms, but can also dis-
cover the new problems. The principles of digital watermarking, steganography,
and forensics in this book are illustrated with plentiful graphs and examples in
order to simplify the problems, so readers can easily understand even complicated
theories. Several robust algorithms that are presented in this book to illustrate the
framework provide assistance and tools in understanding and implementing the
fundamental principles.
Overview of the Book

The book is divided into four parts: Multimedia Mining and Classification (Chapters
1 through 3), Watermarking (Chapters 4 through 10), Steganography (Chapters 11
through 14), and Forensics (Chapters 15 through 20). In Chapter 1, the feasibility,
techniques, and demonstrations of discovering hidden knowledge by applying multi-
media duplicate mining methods to the massive multimedia content are introduced.
Three promising knowledge-discovery applications are demonstrated to show the
benefits of duplicate mining. In Chapter 2, a new framework is proposed for video
concept classification with the help of discriminative learning and multiple
correspondence analysis. In Chapter 3, an improved bag of feature and feature
vocabulary-based technique of replacing the Harris-affine detection method by a
ix
x Preface
random sampling procedure together with an increased number of sample points is

presented. It is shown that the method improves categorization accuracy on a five-
category problem using the Caltech-4 dataset.
Chapters 4 through 10 address digital watermarking techniques. In Chapter 4,
the problems of detecting visible watermarks from images and removing the
watermarks to generate watermark-free images are described. A new Fourier-
based image alignment method with iterative refinement for the watermark detec-
tion is presented. In Chapter 5, a digital watermarking technique based on a chaotic
map and a reference register is presented. A block-based chaotic map, which out-
performs the traditional one by breaking local spatial similarity, is used to increase
the amount of significant coefficients in the transformed image. A reference regis-
ter is then employed to locate specific coefficients of a container efficiently for
watermark embedding and extraction. In Chapter 6, a pseudo-random pixel rear-
rangement algorithm to improve the security of most image watermarking tech-
niques is introduced. It rearranges image pixels based on the properties of Gaussian
integers and results in a more random-looking image transformation that, in turn,
significantly improves the security of the embedded watermark. In Chapter 7, an
overview of existing fragile reversible data-hiding schemes is provided. Basic
schemes are introduced and examples are illustrated to explain some complicated
schemes. In Chapter 8, a novel semi-fragile spatial watermarking method based on
local binary pattern (LBP) operators by using local pixel contrast for the embed-
ding and extraction of watermarks is presented. A general framework for multi-
level image watermarking is also extended. In Chapter 9, an efficient authentication
method for JPEG images based on genetic algorithms (GA) is presented. A two-
level detection strategy is also introduced to reduce the false acceptance ratio of
invalid blocks. In Chapter 10, an efficient block-based fragile watermarking sys-
tem for tamper localization and recovery of images is proposed. The cyclic redun-
dancy checksum (CRC) is used to authenticate the feature of a block stored in a
pair of mapping blocks. The proposed scheme localizes the tampered area irre-
spective of the gray value, and the detection strategy is designed in such a way to
reduce false alarms.
Chapters 11 through 14 cover digital steganography techniques. Chapter 11 sur-
veys image steganography and steganalysis. It introduces the key concepts behind
image steganography and steganalysis, the history and origin of steganography,
and the steganography and steganalysis tools currently available. In Chapter 12,
different types of image steganographic schemes based on vector quantization
(VQ) are presented. The concept of VQ is introduced first, and the main compo-
nents, such as the input image to be compressed, the VQ codebook, the VQ indi-
ces of input image, and the reconstructed image, are applied to digital
steganography. In Chapter 13, a different evolutionary approach, named differen-
tial evolution (DE), is used to increase the performance of the steganographic
system. The key element that distinguishes DE from other population-based
approaches is differential mutation, which aims to find the global optimum of a
multidimensional, multimodal function. In Chapter 14, a robust steganographic
system is presented by artificially counterfeiting statistic features instead of the
traditional strategy by avoiding the change of statistic features. A GA-based
Preface xi
methodology by adjusting gray values of a cover-image while creating the desired

statistic features to generate the stego-images that can break the inspection of
steganalytic systems is described.
Chapters 15 through 20 discuss digital forensic techniques. In Chapter 15, a
robust image inpainting algorithm to determine a new filling order based on the
structure priority value is introduced. The algorithm can fill the missing area with
more accurate structure information. In addition, the dynamic searching range is
used to improve the efficiency of finding the best patches in the source region. In
Chapter 16, the techniques of copy–cover image forgery are discussed and four
detection methods for copy–cover forgery detection, which are based on PCA,
DCT, spatial domain, and statistical domain, are compared. Their effectiveness
and sensitivity under the influences of Gaussian blurring and lossy JPEG com-
pressions are investigated. In Chapter 17, an algorithm for both modification
detection and localization, whose structure can support parallel-processing mode,
is proposed. The mechanism of both changeable-parameter and self-synchroniza-
tion is used to achieve all the performance requirements of hash function. In
Chapter 18, video forensics is discussed, in which a major proportion of related
research work is to perform mining for criminal evidence in videos recorded by a
heterogeneous collection of surveillance camcorders. This is a new interdisciplin-
ary field, and people working in the field need video-processing skills as well as
an in-depth knowledge of forensic science; hence, the barrier for entering the field
is high. In Chapter 19, the problems of vulnerability in the original multi-chaotic
systems–based image encryption scheme are analyzed. A self-synchronizing
method is proposed as an enhancement measure to solve the problems and defeat
cryptanalysis. In Chapter 20, recent advances on human motion behavior model-
ing, including abilities to detect changes in human motion behavior, and to clas-
sify the types of small group motion behavior, are presented. Experimental results
on several public human action datasets are provided to demonstrate the effec-
tiveness of the presented methods in addressing these human behavior modeling
challenges.
Features of the Book

• New state-of-the-art techniques for digital watermarking, steganography,
and forensics
• Numerous practical examples
• A more intuitive development to the complex technology
• Updated bibliography
• Extensive discussion on watermarking, steganography, and forensics
• Inclusion of watermarking, steganalysis, and forensics techniques and their
counter-examples
Feedback on the Book

It is my hope that an opportunity is given to correct any error in this book; therefore,
please provide a clear description of any error that you may find.
xii Preface
Your suggestions on how to improve the textbook are always welcome. For this,
use either email ([email protected]) or regular mail to the author:
Dr. Frank Y. Shih
College of Computing Sciences
New Jersey Institute of Technology
University Heights
Newark, NJ 07102-1982
Acknowledgments
My most heartfelt thanks and all of my love go to my wife and two children for their
great encouragement, patience, and sacrifice in allowing me to work at all hours
from home. I would like to thank my parents for educating me from early childhood
to motivate my writing and realize my own potential.
I am very lucky to have world-renown professors and researchers contribute their
experiences in the book chapters. They ended up writing almost half of this book
and improving the book’s overall quality by several folds. I also want to thank the
editors who helped with the book production and advised on sections that needed
clarification.
Portions of the book appeared in earlier forms as conference papers, journal
papers, or theses by my students at the New Jersey Institute of Technology. Additionally,
portions of the book are contributed by various experts in the fields.
I gratefully acknowledge the following publishers for giving me permission to
reuse texts and figures that appeared in some of these earlier publications: the
Institute of Electrical and Electronic Engineers (IEEE), Bentham Open, Elsevier,
Springer, and Wiley Publishers.
xiii
Editor
Frank Y. Shih received his BS degree from the National Cheng Kung University,
Tainan, Taiwan, in 1980, an MS degree from the State University of New York,
Stony Brook, USA, in 1983, and a PhD degree from Purdue University, West
Lafayette, Indiana, USA, in 1987. He is currently a professor jointly appointed in the
Department of Computer Science, the Department of Electrical and Computer
Engineering, and the Department of Biomedical Engineering at New Jersey Institute
of Technology, Newark, New Jersey. He is also director of the Computer Vision
Laboratory, New Jersey Institute of Technology, Newark, New Jersey. Dr. Shih has
held visiting professor positions at Princeton University; Columbia University;
National Taiwan University; National Institute of Informatics, Tokyo; Conservatoire
National Des Arts Et Metiers, Paris; and Central South University, Changsha, China.
He is an internationally renowned scholar and serves as the editor-in-chief for
International Journal of Multimedia Intelligence and Security (IJMIS). Dr. Shih is
currently on the editorial board of Pattern Recognition, Pattern Recognition Letters,
International Journal of Pattern Recognition and Artificial Intelligence, Journal of
Information Hiding and Multimedia Signal Processing, Recent Patents on
Engineering, Recent Patents on Computer Science, The Open Nanoscience Journal,
International Journal of Internet Protocol Technology, Journal of Internet
Technology, ISRN Signal Processing, and ISRN Machine Vision. He has served as a
steering member, committee member, and session chair for numerous professional
conferences and workshops. Dr. Shih has received numerous grants from the National
Science Foundation, Navy and Air Force, and industry. He was the recipient of the
Research Initiation Award of the NSF in 1991 and the Board of Overseers Excellence
in Research Award of NJIT in 2009.
Dr. Shih has made significant contributions to information hiding, focusing on the
security and robustness of digital watermarking and steganography. He has authored
three books: Digital Watermarking and Steganography, Image Processing and
Mathematical Morphology, and Image Processing and Pattern Recognition. He has
published 115 journal papers, 95 conference papers, and 10 book chapters. His cur-
rent research interests include image processing, computer vision, watermarking and
steganography, digital forensics, sensor networks, pattern recognition, bioinformat-
ics, information security, robotics, fuzzy logic, and neural networks.
xv
Contributors
Mayra Bachrach received her MS in computer science from New Jersey Institute of
Technology in 2011. She is a teacher at Bergen County Academies, Hackensack,
New Jersey. Her research interests include image processing and databases.
Chin-Chen Chang received his BS in applied mathematics in 1977 and MS in com-
puter and decision sciences in 1979, both from the National Tsing Hua University,
Hsinchu, Taiwan. He received his PhD in computer engineering in 1982 from the
National Chiao Tung University, Hsinchu, Taiwan. During the academic years of
1980–1983, he was on the faculty of the Department of Computer Engineering at the
National Chiao Tung University. From 1983 to 1989, he was on the faculty of the
Institute of Applied Mathematics, National Chung Hsing University, Taichung,
Taiwan. From August 1989 to July 1992, he was the head of, and a professor in, the
Institute of Computer Science and Information Engineering at the National Chung
Cheng University, Chiayi, Taiwan. From August 1992 to July 1995, he was the dean
of the College of Engineering at the same university. From August 1995 to October
1997, he was the provost at the National Chung Cheng University. From September
1996 to October 1997, Dr. Chang was the acting president at the National Chung
Cheng University. From July 1998 to June 2000, he was the director of the Advisory
Office of the Ministry of Education of the People’s Republic of China. From 2002 to
2005, he was a chair professor of National Chung Cheng University. Since February
2005, he has been a chair professor of Feng Chia University. In addition, he has
served as a consultant to several research institutes and government departments.
His current research interests include database design, computer cryptography,
image compression, and data structures.
I-Cheng Chang received a BS in nuclear engineering in 1987, and an MS and PhD
in electrical engineering in 1991 and 1999, respectively, all from National Tsing Hua
University, Hsinchu, Taiwan. In 1999, he joined Opto-Electronics & Systems
Laboratories, Industrial Technology Research Institute, Hsinchu, Taiwan, as an engi-
neer and project leader. He is currently an associate professor in the Department of
Computer Science and Information Engineering, National Dong Hwa University,
Hualien, Taiwan. His research interests include video and image processing, surveil-
lance network system, computer vision, and multimedia system design. Dr. Chang
received the Annual Best Paper Award from the Journal of Information Science and
Engineering in 2002, and research awards from the Industrial Technology Research
Institute in 2002 and 2003. He is a member of the IEEE, and the IPPR of Taiwan,
People’s Republic of China.
Shu-Ching Chen has been a full professor in the School of Computing and
Information Sciences (SCIS), Florida International University (FIU), Miami since
August 2009. Prior to that, he was an assistant/associate professor in SCIS at FIU
from 1999. He received master’s degrees in computer science, electrical engineering,
and civil engineering in 1992, 1995, and 1996, respectively, and a PhD in electrical
xvii
xviii Contributors
and computer engineering in 1998, all from Purdue University, West Lafayette,
Indiana, USA. His main research interests include content-based image/video
retrieval, distributed multimedia database management systems, multimedia data
mining, multimedia systems, and disaster information management. Dr. Chen has
authored or coauthored more than 240 research papers in journals, refereed confer-
ence/symposium/workshop proceedings, book chapters, and one book. Dr. Chen
received the best paper award from the 2006 IEEE International Symposium on
Multimedia. He was awarded the IEEE Systems, Man, and Cybernetics (SMC)
Society’s Outstanding Contribution Award in 2005 and was the co-recipient of the
IEEE Most Active SMC Technical Committee Award in 2006. He was also awarded
the Inaugural Excellence in Graduate Mentorship Award from FIU in 2006, the
University Outstanding Faculty Research Award from FIU in 2004, the Excellence
in Mentorship Award from SCIS in 2010, the Outstanding Faculty Service Award
from SCIS in 2004, and the Outstanding Faculty Research Award from SCIS in
2002. He is the chair of IEEE Computer Society Technical Committee on Multimedia
Computing and co-chair of IEEE Systems, Man, and Cybernetics Society’s Technical
Committee on Knowledge Acquisition in Intelligent Systems. He is a fellow of
Society for Information Reuse and Integration (SIRI).
Ya-Yun Cheng received a BS from the University of Kaohsiung, Taiwan, in 2010.
She is currently a master’s student in the Institute of Information Systems and
Applications at National Tsing Hua University. Her research interest includes com-
puter vision, machine learning, medical image analysis, and image segmentation.
Venkata Gopal Edupuganti received Bachelor of Technology and Master of
Technology degrees in Information Technology from the University of Hyderabad,
Hyderabad, India, in 2004 and 2007, respectively. He is currently working toward his
PhD in the Department of Computer Science, New Jersey Institute of Technology,
Newark, New Jersey. His research interests include image processing, image retrieval,
watermarking, and pattern recognition.
Chia-We Hsu received her BS in computer science and information engineering from
Da Yeh University, Changhua, Taiwan, in 2004, and her MS in computer science and
information engineering from Dong Hwa University, Hualien, Taiwan, in 2007. Her
research interests include image/video processing and pattern recognition.
Aleksey Koval received his PhD in computer science from the New Jersey Institute
of Technology (NJIT) in 2011. His primary field of study is cryptography and secu-
rity. Dr. Koval has vast work experience in the IT industry working for AT&T and
IBM. He currently works for IBM.
Shang-Hong Lai received his BS and MS in electrical engineering from the National
Tsing Hua University, Hsinchu, Taiwan, and a PhD in electrical and computer engi-
neering from the University of Florida, Gainesville, in 1986, 1988, and 1995, respec-
tively. He joined Siemens Corporate Research in Princeton, New Jersey, as a member
of the technical staff in 1995. In 1999, he became a faculty member in the Department
of Computer Science, National Tsing Hua University, Taiwan. He is currently a pro-
fessor in the same department and the director of the computer and communication
Contributors xix
center in the university. In 2004, he was a visiting scholar at Princeton University.

Dr. Lai’s research interests include computer vision, visual computing, pattern rec-
ognition, medical imaging, and multimedia signal processing. He has authored more
than 180 papers published in related international journals and conferences. He holds
many patents for inventions related to computer vision and medical image analysis.
He has been a program committee member of several international conferences,
including CVPR, ICCV, ECCV, ACCV, ICPR, and ICME.
Hong-Yuan Mark Liao received his BS in physics from National Tsing-Hua
University, Hsin-Chu, Taiwan, in 1981, and an MS and PhD in electrical engineering
from Northwestern University in 1985 and 1990, respectively. In July 1991, he joined
the Institute of Information Science, Academia Sinica, Taiwan. Currently he is a
research fellow. In 2008, he became the division chair of the Computer Science and
Information Engineering Division II, National Science Council of Taiwan. He is
jointly appointed as a professor in the Computer Science and Information Engineering
Department of National Chiao-Tung University and the Department of Electrical
Engineering of National Cheng Kung University. Since February 2009, he has been
jointly appointed as the multimedia information chair professor of National Chung
Hsing University. In August 2010, he was appointed as an adjunct chair professor of
Chung Yuan Christian University. His current research interests include multimedia
signal processing, video-based surveillance systems, video forensics, and multime-
dia protection.
Dr. Liao was a recipient of the Young Investigators’ award from Academia Sinica
in 1998. He received the Distinguished Research Award from the National Science
Council of Taiwan in 2003 and 2010, and the National Invention Award of Taiwan in
2004. In 2008, he received a Distinguished Scholar Research Project Award from
National Science Council of Taiwan. In 2010, he received the Academia Sinica
Investigator Award. In June 2004, he served as the conference co-chair of the 5th
International Conference on Multimedia and Exposition (ICME) and technical co-
chair of the 8th ICME held in Beijing. From 2006 to 2008, Dr. Liao was the president
of the Image Processing and Pattern Recognition Society of Taiwan.
Dr. Liao is on the editorial boards of the IEEE Signal Processing Magazine, IEEE
Transactions on Image Processing, and IEEE Transactions on Information Forensics
and Security. He was an associate editor of IEEE Transactions on Multimedia dur-
ing 1998–2001.
Xiaofeng Liao received his PhD from the University of Electronic Science and
Technology of China in 1997. He is a professor in the Department of Computer
Science and Engineering, Chongqing University, Chongqing, China. His research
interests include information security, image processing, neural networks, and artifi-
cial intelligence.
Shinfeng D. Lin received his BS in automatic control engineering from Feng Chia
University, Taichung, Taiwan, in 1980, and an ME and PhD in electrical engineering
from Mississippi State University in 1985 and 1991, respectively. He is currently a
professor, chairman, and deputy dean in the Department of Computer Science and
Information Engineering, College of Science and Engineering, National Dong Hwa
xx Contributors
University, Taiwan. Dr. Lin was the director of the Bureau of Education, Hualien
County, Taiwan, from January 2002 to September 2003. His research interests
include signal/image processing, image/video compression, and information secu-
rity. He won the Gold Medal Award at the 2005 International Trade Fair “Ideas–
Inventions–New Products” (IENA), Nuremberg, Germany. Dr. Lin is a member of
Tau Beta Pi and Eta Kappa Nu.
Hong Man received his PhD in electrical engineering from Georgia Institute of
Technology in 1999. He joined the Stevens Institute of Technology in 2000. He is
currently an associate professor in the Electrical and Computer Engineering
Department, the director of the Visual Information Environment Laboratory, and the
director of the undergraduate computer engineering program at Stevens. He served
on the Organizing Committees of IEEE ICME (2007 and 2009), IEEE MMSP (2002
and 2005), and the Technical Program Committees of various IEEE conferences. He
is a member of the IEEE Technical Committees on Multimedia (MMTC) and
Communications and Information Security (CISTC). His research interests include
signal and image processing, pattern recognition and data mining, on which he has
published more than 70 technical journal and conference papers. He is a senior mem-
ber of IEEE and a member of ACM, ASEE.
Sebastien Poullot obtained his PhD in computer science in 2009 after earning two
master’s degrees in bio-informatics (2004) and computer sciences (2005). He was
invited by Professor Satoh’s team at the National Institute of Informatics, and later
worked as a project researcher (2009–2011). Since 2011, he has been employed at
INRIA Rocquencourt in the IMEDIA team. His research is mainly focused on video
and image retrieval, especially the scalability of applications.
Shin’ichi Satoh received his BE in electronics engineering in 1987, and his ME and
PhD in information engineering in 1989 and 1992 at the University of Tokyo. He
joined the National Center for Science Information Systems (NACSIS), Tokyo, in
1992. He has been a full professor at the National Institute of Informatics (NII),
Tokyo, since 2004. He was a visiting scientist at the Robotics Institute, Carnegie
Mellon University, from 1995 to 1997. His research interests include image process-
ing, video content analysis, and multimedia database. Currently, he is leading the
video-processing project at NII.
Alexander Sheppard received a BS in computer science from the New Jersey

Institute of Technology. He is currently working toward a PhD in computer science
at the University of North Carolina, Chapel Hill. His research interests include image
processing and pattern recognition.
Shih-Chieh Shie received his BS from the Department of Computer Science and
Engineering, Tatung Institute of Technology, Taipei, Taiwan, People’s Republic of
China, in 1996, and an MS and PhD from the Department of Computer Science and
Information Engineering, National Dong Hwa University, Hualien, Taiwan, People’s
Republic of China, in 2000 and 2005, respectively. Dr. Shie is currently an assistant
Contributors xxi
professor in the Department of Computer Science and Information Engineering,

National Formosa University, Taiwan, People’s Republic of China. His research inter-
ests include information hiding, image processing, image compression, and image
watermarking.
Mei-Ling Shyu has been an associate professor in the Department of Electrical and
Computer Engineering (ECE), University of Miami (UM), since June 2005. Prior to
that, she was an assistant professor in ECE at UM dating from January 2000. She
received her PhD from the School of Electrical and Computer Engineering, Purdue
University, West Lafayette, Indiana, USA in 1999 and three master’s degrees in com-
puter science, electrical engineering, and restaurant, hotel, institutional, and tourism
management from Purdue University in 1992, 1995, and 1997. Her research interests
include multimedia data mining, multimedia information management and retrieval,
and network security. She has coauthored more than 200 technical papers published
in prestigious journals, book chapters, and refereed conference/workshop/sympo-
sium proceedings. She is the vice chair of the IEEE Computer Society Technical
Committee on Multimedia Computing. She received the Best Student Paper Award
with her student at the Third IEEE International Conference on Semantic Computing
in September 2009, the Best Published Journal Article in International Journal of
Multimedia Data Engineering and Management (IJMDEM) for 2010 Award from
IGI Global in April 2011, and the Johnson A. Edosomwan Scholarly Productivity
Award from the College of Engineering at UM in 2007. She is a fellow of SIRI.
Hong-Ren Su received his BS in biomedical imaging and radiological science from

the National Yang Ming University, Taiwan, in 2000 and an MS in nuclear science
from the National Tsing Hua University, Taiwan, in 2002. He is currently working
toward obtaining a PhD from the Institute of Information Systems and Applications
at National Tsing Hua University. He is a student member of the IEEE. His research
interests includes computer vision, machine learning, and medical image analysis.
His current address is Institute of Systems and Applications, National Tsing Hua
University, Hsinchu, Taiwan.
Ming-Ting Sun received his BS from the National Taiwan University in 1976, an
MS from the University of Texas at Arlington in 1981, and a PhD from the University
of California, Los Angeles in 1985, all in electrical engineering. Dr. Sun joined the
faculty of the University of Washington in September 1996. Before that, he was the
director of the Video Signal Processing Group at Bellcore (now Telcordia). He has
been a chair professor at Tsing Hwa University, and a visiting professor at Tokyo
University, National Taiwan University, and National Chung-Cheng University. His
research interests include machine learning and video processing. Dr. Sun has been
awarded 11 patents and has published more than 200 technical publications, includ-
ing 15 book chapters in the area of video technology. He was actively involved in the
development of H.261, MPEG-1, and MPEG-2 video-coding standards. He has also
coedited a book titled Compressed Video Over Networks. He was the editor-in-chief
of IEEE Transactions on Multimedia during 2000–2001. He received an IEEE
CASS Golden Jubilee Medal in 2000, and he was a co-chair of the SPIE VCIP
xxii Contributors
(Visual Communication and Image Processing) 2000 Conference. He was the editor-
in-chief of IEEE Transactions on Circuits and Systems for Video Technology
(T-CSVT) during 1995–1997 and the Express Letter editor of T-CSVT during 1993–
1994. He was a corecipient of the T-CSVT Best Paper Award in 1993. From 1988 to
1991, he served as the chairman of the IEEE CAS Standards Committee and estab-
lished an IEEE Inverse Discrete Cosine Transform Standard. He received an Award
of Excellence from Bellcore in 1987 for his work on the digital subscriber line.
Dr. Sun was elected as a Fellow of the IEEE in 1996.
Hsiao-Rong Tyan received her BS in electronic engineering from Chung Yuan

Christian University, Chungli, Taiwan, in 1984, and an MS and PhD in computer
science from Northwestern University, Evanston, Illinois, in 1987 and 1992, respec-
tively. She is an associate professor in the Department of Information and Computer
Engineering, Chung Yuan Christian University, Chungli, Taiwan, where she cur-
rently conducts research in the areas of computer security, computer networks, and
real-time intelligent systems.
Boris S. Verkhovsky is a professor in the Computer Science Department at the New

Jersey Institute of Technology (NJIT). He received his PhD in computer science
jointly from Latvia State University and from the Academy of Sciences of the USSR
(Central Institute of Mathematics and Economics, Moscow). From his prior affilia-
tions at the Scientific Research Institute of Computers (Moscow), Princeton
University, IBM Thomas J. Watson Research Center, Bell Laboratories and since
1986 at the NJIT, he acquired research interests and experience in cryptography,
communication security, large-scale systems optimal design and control, optimiza-
tion, and algorithm design.
For the last 20 years, Dr. Verkhovsky’s research activity has been centered on the
design and analysis of cryptographic systems and information assurance algorithms.
Prior to NJIT he worked at the Academy of Sciences of the USSR, IBM Thomas J.
Watson Research Center, was associate professor at Princeton University, a member
of Technical Staff at Bell Labs, and was Endowed Chair Professor, University of
Colorado. Professor Verkhovsky is listed in Marquis Who’s Who in America. Professor
Verkhovsky is editor-in-chief of the International Journal of Communications,
Network and System Sciences.
Zhi-Hui Wang received a BS in software engineering in 2004 from the North

Eastern University, Shenyang, China. She received an MS in software engineering in
2007 and a PhD in computer software and theory in 2010, both from the Dalian
University of Technology, Dalian, China. Since November 2011, she has been a visit-
ing scholar of the University of Washington. Her current research interests include
information hiding and image compression.
Xiaomeng Wu is a project researcher at the National Institute of Informatics, Japan.

He graduated from the University of Shanghai for Science and Technology with a
baccalaureate degree in thermal energy engineering in 2001. He graduated from the
University of Tokyo with a doctoral degree in information and communication engi-
neering in 2007. His research interests include multimedia indexing, multimedia
Contributors xxiii
retrieval, and near-duplicate detection. He is a member of the IEEE Computer

Society and ACM.
Yi-Ta Wu received his BS in physics from Tamkang University, Taipei, Taiwan, in

1995, and an MS in computer science from National Dong-Hwa University, Hualien,
Taiwan, in 1997. Dr. Wu received his PhD from the Department of Computer Science,
New Jersey Institute of Technology, in May 2005. He was a research fellow at
the University of Michigan, Ann Arbor, and is currently affiliated with Industrial
Technology Research Institute, Taiwan. His research interests include image/video
processing, mathematical morphology, image watermarking, steganography, surveil-
lance system, robot vision, pattern recognition, shortest path planning, and artificial
intelligence.
Di Xiao received a BS from Sichuan University, Chengdu, China in 1997. He

obtained both an MS and PhD from Chongqing University, Chongqing, China, in
2002 and 2005, respectively. He is currently a professor in the College of Computer
Science and Engineering, Chongqing University, Chongqing, China. His research
interests include chaotic dynamical systems-based cryptography, information
authentication, digital watermark, and digital image forensics.
Yafeng Yin is currently a PhD candidate in the Electrical and Computer Engineering
Department at Stevens Institute of Technology, and a research intern at Bell
Laboratories, Alcatel-Lucent. He received his bachelor’s from the Department of
Automatic Control at Beijing Institute of Technology in 2005. His research interests
include image processing, pattern recognition and machine learning, with applications
to video-based object recognition, tracking, and human group behavior analysis.
Yuan Yuan obtained both an BS and MS from the Department of Computer Science
and Technology, Zhejiang University of Technology, Hangzhou, People’s Republic
of China, in 2004 and 2006, respectively. He secured a PhD in computer science
from the New Jersey Institute of Technology, Newark, New Jersey, in 2011. He is
currently affiliated with Amazon Corporation, Seattle, Washington, DC.
Wenyin Zhang is an assistant professor in the School of Informatics, Linyi

University, Linyi, Shandong, People’s Republic of China. He obtained his master’s in
2002 from Shandong University of Science and Technologies and was awarded a
doctorate by the Chinese Academy of Sciences in 2005. His research interest includes
image processing, information hiding, digital watermarking.
Qiusha Zhu is currently a PhD candidate at the Department of Electrical and
Computer Engineering (ECE), University of Miami. She received her MS in
electronic engineering from School of Electronic, Information and Electrical
Engineering, Shanghai Jiao Tong University in March 2009 and a BS in electronic
engineering from Shanghai University as an outstanding graduate in June 2006.
She worked as a database apprentice at DOW Chemical, Shanghai from November
2007 to June 2008. Her research interests include multimedia information retrieval
and data mining, especially how to effectively retrieve online images and videos, and
how to fuse information available from different sources to boost this process.
Part I
Multimedia Mining and
Classification
1 Multimedia Duplicate
Mining toward
Knowledge Discovery
Xiaomeng Wu, Sebastien Poullot,
and Shin’ichi Satoh
Contents
1.1 Introduction.......................................................................................................4
1.2 Selection Criterion of Duplicate Mining Methods............................................5
1.2.1 Exact Duplicate Mining.........................................................................5
1.2.2 Near-Duplicate Mining..........................................................................6
1.3 TV Commercial Mining for Sociological Analysis...........................................7
1.3.1 Background............................................................................................7
1.3.2 Temporal Recurrence Hashing Algorithm............................................8
1.3.3 Knowledge Discovery Based on CF Mining.........................................9
1.4 News Story Retrieval and Threading.............................................................. 12
1.4.1 Background.......................................................................................... 12
1.4.2 StoryRank............................................................................................ 14
1.4.3 Experimentation Evaluation................................................................ 16
1.5 Indexing for Scaling up Video Mining............................................................ 17
1.5.1 Background.......................................................................................... 17
1.5.2 Glocal Description............................................................................... 18
1.5.3 Cross-Dimensional Indexing............................................................... 19
1.5.4 Shape Embedding................................................................................ 21
1.5.5 Temporal Consistency.......................................................................... 23
1.5.6 Experiments and Results..................................................................... 23
1.5.6.1 Quality.................................................................................. 23
1.5.6.2 Scalability.............................................................................25
1.5.7 Conclusions..........................................................................................26
1.6 Conclusions and Future Issues.........................................................................26
Acknowledgment...................................................................................................... 27
Terminology Indexes................................................................................................. 27
References................................................................................................................. 27
3
4 Multimedia Security
1.1 Introduction
The spread of digital multimedia content and services in the field of broadcasting
and on the Internet has made multimedia data mining an important technology for
transforming these sources into business intelligence for content owners, publishers,
and distributors. A recent research domain known as multimedia duplicate mining
(MDM) has emerged largely in response to this technological trend. The “multime-
dia duplicate mining” domain is based on detecting image, video, or audio copies
from a test collection of multimedia resources. One very rich area of application is
digital rights management, where the unauthorized or prohibited use of digital media
on file-sharing networks can be detected to avoid copyright violations. The primary
thesis of MDM in this application is “the media itself is the watermark,” that is, the
media (image, video, or audio) contains enough unique information to be used to
detect copies (Hampapur et al., 2002). The key advantage of MDM over other
technologies, for example, the watermarking, is the fact that it can be introduced
after copies are made and can be applied to content that is already in circulation.
Monitoring commercial films (CFs) is an important and valuable task for
competitive marketing analysis, for advertising planning, and as a barometer of the
advertising industry’s health in the field of market research (Li et al., 2005; Gauch
and Shivadas, 2006; Herley, 2006; Berrani et al., 2008; Dohring and Lienhart, 2009;
Putpuek et al., 2010; Wu and Satoh, 2011). In the field of broadcast media research,
duplicate videos shared by multiple news programs imply that there are latent seman-
tic relations between news stories. This information can be used to define the
similarities between news stories; thus, it is useful for news story tracking, threading,
and ranking (Duygulu et al., 2004; Zhai and Shah, 2005; Wu et al., 2008a; Wu et al.,
2010). From another viewpoint, duplicate videos play a critical role in assessing the
novelty and redundancy among news stories, and can help in identifying any fresh
development among a huge volume of information in a limited amount of time (Wu
et al., 2007a; Wu et al., 2008b). Additionally, MDM can be used to detect filler
materials, for example, opening CG shots, anchor person shots, and weather charts
in television, or background music in radio broadcasting (Satoh, 2002).
This chapter discusses the feasibility, techniques, and demonstrations of discover-
ing hidden knowledge by applying MDM methods to the massive amount of
multimedia content. We start by discussing the requirements and selection criteria
for the duplicate mining methods in terms of the accuracy and scalability. These
claims involve the sampling and description of videos, the indexing structure, and
the retrieval process, which depend on the application purposes. We introduce three
promising knowledge-discovery applications to show the benefits of duplicate
mining. The first application (Wu and Satoh, 2011) is dedicated to fully unsupervised
TV commercial mining for sociological analysis. It uses a dual-stage temporal
recurrence hashing algorithm for ultra-fast detection of identical video sequences.
The second application (Wu et al., 2010) focuses on news story retrieval and
threading: it uses a one-to-one symmetric algorithm with a local interest point index
structure to accurately detect identical news events. The third application (Poullot
et al., 2008, 2009) is for large-scale cross-domain video mining. It exploits any
weak geometric consistencies between near-duplicate images and addresses the
Multimedia Duplicate Mining toward Knowledge Discovery 5
s calability issue of a near-duplicate search. Finally, a discussion on these techniques

and applications is given.
1.2 Selection Criterion of Duplicate Mining Methods

Choosing a duplicate mining method can be difficult because of the variety of methods
currently available. The application provider must decide which method is best suited
to their individual needs, and, of course, the type of duplicate that they want to use as
the target. In this sense, the definition of a duplicate is generally subjective and, to a
certain extent, does depend on the type of application being taken into consideration.
The application ranges from exact duplicate mining, where no changes are allowed,
to a more general definition that requires the resources to be of the same scene, but
with possibly strong photometric or geometric transformations.
1.2.1 Exact Duplicate Mining

One direction for MDM is to mine duplicate videos or audios derived from the original
resource without any or with very few transformations. This type of duplicate is known
as an exact duplicate or exact copy. Typical cases include TV CFs and file footages
used in news. Most existing studies on exact duplicate mining introduce the concept of
fingerprint or “hash” functions. This is a signal that has been extracted from each
sample of a video. It uniquely identifies the sample as genuine and cannot be decoded
to get the original one, because the fingerprint is made by some of the features of the
sample.
The fingerprints are computed on the sampled frames given the video dataset in
order to construct a hash table, into which the frames are inserted by regarding the
corresponding fingerprint as an inverted index. An important assumption made here
is that identical or approximately identical frames of all the video duplicates have
exactly the same fingerprint, so a hash collision occurs in the corresponding hash
bucket. Based on this assumption, approximately identical frames of all the exact
duplicates can be efficiently retrieved by using a collision attack process.
In most cases, the sampling rate, which defines the number of sampled frames per
unit of time (usually seconds) taken from a continuous video to make a discrete
fingerprint, is as dense as the allowed maximal frame frequency of the video, for
example, 29.97 (NTSC) or 25 (PAL) frames per second for an MPEG-2 video. This
high-density sampling can accommodate the fingerprinting errors, that is, different
fingerprints derived from identical frames, and allows the detection to be precise in
order to locate the boundaries of the duplicates. The description of the video, based
on which the fingerprint is extracted, can be color histogram (Li et al., 2005), ordinal
signature (Li et al., 2005), color moments (Gauch and Shivadas, 2006), gradient
histogram (Dohring and Lienhart, 2009), luminance moments (Wu and Satoh, 2011),
and acoustic signatures (Haitsma and Kalker, 2002; Cano et al., 2005). All these
global descriptors have been effectively demonstrated, thanks to the limited number
of transformations between the original resource and the duplicate. Therefore,
most researchers attach more importance to the efficiency and scalability than the
effectiveness of the exact duplicate mining methods.
1.2.2 Near-Duplicate Mining
Near duplicates are considered the slight alterations of videos or images issued
from the same original source. In other literature, near duplicates are also known as
copies. The alterations are obtained from an original using various photometric or
geometric transformations, for example, cam cording, picture in picture, insertions
of patterns, strong reencoding, gamma changes, decreases in quality, postproduction
transformations, and so on. There is significant diversity in the nature and amplitude
of the encountered transformations. Near duplicates are widespread among the
current worldwide video pool. A typical example is the digital copies of copyrighted
materials transferred on file-sharing networks or video-hosting services, for exam-
ple, Flickr, YouTube, and so on. Two examples are shown in Figure 1.1. The
transformation of the example on the left is cam-cording, and those of the right one
include a change of contrast, blurring, noising, and scaling.
Near-duplicate mining is considered a more difficult case than exact duplicate
mining because the videos or images can be strongly transformed. The assumption
of identical or approximately identical images from among all the duplicates with
exactly the same fingerprint (Section 1.2.1) becomes too weak to make when
detecting near duplicates. Most existing studies on near-duplicate mining use an
image distance measure instead of the fingerprinting-based hash collision attack
process for finding approximately identical images. An image distance measure
compares the similarities between two images in various dimensions, for example,
the color, texture, and shape. Most of these studies are designed for images, but can
be extended to the video (an image stream) domain by embedding the temporal
consistency constraints between duplicate images in the mining process.
A general representation of the images must be chosen to compute the similarities
between two images. A picture may be represented by a single global feature (e.g., color
histogram, edge orientation histogram, Fourier, or Hough) or by many small local ones
(e.g., SIFT, Harris, or dipoles: Gouet and Boujemaa, 2001; Lowe, 2004; Mikolajczyk et al.,
2005; Joly et al., 2007). The main difference consists in the precision of the description
and in the scalability of the approach, two inversely proportional parameters. Methods
with global features usually have a higher efficiency than those with local ones. However,
local descriptions have recently become more popular because of their superiority over the
global ones in terms of their discriminative power, especially in the most difficult cases
where the duplicate videos are short and strongly transformed (Law-To et al., 2007).
Given a video dataset, the local descriptors are computed on the sampled frames
in order to construct a descriptor database. Most methods choose a sampling rate
Figure 1.1 Two examples of near duplicates. (From NISV, Sound and Vision Video
Collection 2008. With permission.)
(e.g., one frame per second) that is much lower than the allowed maximal frame
frequency of the video to ease the potentially high computation burden of the image
distance measure. The descriptor database can be queried with a video or a picture,
where its local features (Gouet and Boujemaa, 2001; Ke et al., 2004; Jegou et al.,
2008) are used as queries, or be crossed over itself by using each of its features as a
query. A set of candidates is returned for each feature query. A vote is then conducted
to choose the better candidates. Some of these solutions have shown great scalability
concerning the descriptor matching, but the last decision step has a prohibitive cost
when performing mining. In the early 2000s, the bag of words (BoW) representation
(Philbin et al., 2007; Wu et al., 2007b; Yuan et al., 2007) raised up. It was inspired
from the inversed lists used in text search engines. Here, a global representation of
an image is computed by using the local descriptions. A visual vocabulary is first
computed on the sampled local features, by using a K-means algorithm. Then, for a
given image, all its local features are quantized on this vocabulary; the resulting
description is a histogram with a dimensionality equal to the vocabulary size. There
are two major advantages to this approach, the compactness of the feature, and the
suppression of the last decision step. It usually leads to a more scalable solution while
offering good-quality results. One of the notorious disadvantages of BoW is that it
ignores the spatial relationships among the patches, which is very important in image
representation. Therefore, a decision process based on the spatial configuration (e.g.,
with RANSAC) is usually performed as well on the top candidates (Fischler and
Bolles, 1981; Matas and Chum, 2005; Joly et al., 2007).
1.3 TV Commercial Mining for Sociological Analysis

1.3.1 Background
TV commercials, also known as CFs, are a form of communication intended to
persuade an audience to consume or purchase a particular product or service. It is
generally considered the most effective mass-market advertising format of all the
different types of advertising. CF mining has recently attracted the attention of
researchers. This task aims at developing techniques for detecting, localizing, and
identifying CFs broadcast on different channels and in different time slots. CF
removal based on CF detection and localization is an important preprocessing
technique in broadcast media analysis, where CFs are usually considered to be an
annoyance. In market research, monitoring CFs is an important and valuable task for
competitive marketing analysis, for advertising planning, and as a barometer of the
advertising industry’s health. Another promising application of CF mining is to
furnish consumers with a TV CF management system, which categorizes, filters out,
and recommends CFs on the basis of personalized consumer preference.
CF mining techniques can be categorized into knowledge- and repetition-based
techniques. The knowledge-based ones (Albiol et al., 2004; Hua et al., 2005; Duan
et al., 2006; Li et al., 2007) use the intrinsic characteristics of CF (e.g., the presence
of monochrome, silence, or product frames) as prior knowledge for detecting them.
Most knowledge-based techniques are efficient but not generic enough because of
the data-dependent prior knowledge they use. For example, most Japanese TV sta-
tions do not use monochrome or silence frames to separate two consecutive CF

sequences. Other characteristics (e.g., product frame, shot information, and motion
information) normally vary with the nations and regions. Meanwhile, CF can be
regarded as repetition in the stream. Some recent studies (Li et al., 2005; Gauch and
Shivadas, 2006; Herley, 2006; Berrani et al., 2008; Dohring and Lienhart, 2009;
Putpuek et al., 2010) have used this property to detect CFs. Conversely, repetition-
based ones are unsupervised and more generic, but are a large computational bur-
den. In some studies, the required system task is given a test collection of videos
and a reference set of CFs to determine for each CF the location at which the CF
occurs in the test collection. These studies are not fully unsupervised considering
the construction of the reference set. In most cases, the boundaries of the CFs in the
reference set are known. The increase in processing time is linear to that of the size
of the test collection. A more challenging task is, when given a test collection of
videos, how to determine the location, if any, at which some CF occurs. This is also
the main problem that the technique introduced in this section aims to solve. Neither
the reference database nor the CF boundary is provided beforehand. The system
should be fully unsupervised. The algorithm needs to be much more scalable
because the processing time will increase as the square of the test collection size
increases.
1.3.2 Temporal Recurrence Hashing Algorithm

The target is to output a set of CF clusters when given a broadcast video stream as
the input. A CF cluster is a set of multiple identical CF sequences and corresponds to
a specific commercial, for example, an Apple Macintosh commercial. A CF sequence,
which is composed of one or more consecutive fragments, is a long and complete
sequence corresponding to an instance of a commercial. A fragment, which is simi-
lar to but much shorter than a shot or subshot, is a short segment, composed of one
or more consecutive and identical frames. In this study, the visual and audio features
are extracted from all the frames of a broadcast video stream. Consecutive frames
with identical low-level features are assembled into a fragment. An example of the
fragments derived from two recurring CF sequences is shown in Figure 1.2, from
where we can observe a strong temporal consistency among the pairs of recurring
fragments. For instance, the temporal positions of the fragments themselves are con-
secutive, and the temporal intervals between each two fragments are identical to
each other. This temporal consistency is exceptionally useful for distinguishing the
recurring sequences from the nonduplicate ones. The CF mining task can thus
be formulated into searching for recurring fragment pairs with high-temporal
consistency from all the pairs of recurring fragments derived from the entire stream.
In this study, two temporal recurrence fingerprints are proposed for capturing the
temporal information of each recurring fragment pair. The first fingerprint corre-
sponds to the temporal position of the former fragment in the recurring fragment
pair, the resolution of which is set to a given set of minutes so that the consecutive
fragment pairs derived from two recurring CF sequences can be mapped to the same
fingerprint or neighboring ones. The second fingerprint corresponds to the temporal
interval between the two fragments, the resolution of which is set to a given set of
Figure 1.2 Recurring fragment pairs derived from two CF sequences of the same CF cluster.
seconds. As shown in Figure 1.3, all the recurring fragment pairs derived from the
stream are inserted into a 2-dimensional hash table based on the two fingerprints.
By doing so, the recurring fragment pairs with high-temporal consistency can be
automatically assembled into the same bin. These fragments, the assembly of which
corresponds to a pair of long and complete CF sequences, can thus be easily detected
by searching for the hash collisions from the hash table. More details about this
algorithm can be found in Wu and Satoh (2011).
This temporal recurrence hashing algorithm was tested by using a 10-h broadcast
video stream with a manually annotated ground truth for the accuracy evaluation.
The sequence-level accuracy was 98.1%, which is enough to evaluate how precisely
the algorithm can detect and identify recurring CF sequences. The frame-level
accuracy was 97.4%, which evaluates how precisely the algorithm can localize the
start and end positions of the CF sequences in the stream. Three state-of-the-art
studies were implemented for comparison, including two video-based techniques
(Berrani et al., 2008; Dohring and Lienhart, 2009) and one audio-based one (Haitsma
and Kalker, 2002). The temporal recurrence hashing algorithm outperformed the
related studies in terms of both the sequence- and the frame-level accuracy. The
algorithm mined the CFs from the 10-h stream in <4 s, and was more than 10 times
faster than those in the related studies.
1.3.3 Knowledge Discovery Based on CF Mining

CF mining is an important and valuable task for competitive marketing analysis and
is used as a barometer for advertising planning. We discuss a promising application
of CF mining in this section that visualizes the chronological distribution of a
massive amount of CFs and enables the discovery of hidden knowledge in the
advertising market. This application is illustrated in Figure 1.4. It visualizes nine
h1 = 2, h2 = 11
h1 = 6, h2 = 7
h1 = 2, h2 = 4
h1 = 2, h2 = 11
3
h1
4
1 2 3 4 5 6 7 8 9 10 11 12
h2
Figure 1.3 Temporal recurrence hashing.
Figure 1.4 Application of knowledge discovery based on CF mining.

Sat 22 11 10 7 10 14 7 6 6 10 15 4 8 11
Fri 12 4 7 6 6 5 12 12
Day of the week
Thu 13 13 8 3 3 14 4 13
Wed 18 5 10 8 4 3 9 15
Tue 13 9 9 7 6 9 8 15
Mon 10 11 9 3 6 2 4 15
Sun 15 13 14 18 14 14 4 11 5 9 9 11 7 25
00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Hour of the day
Figure 1.5 Asahi Breweries CF chronological distribution.
types of chronological distributions involving five types of information including

broadcast frequency, date, hour of the day, day of the week, and channel. The CFs
were mined from a 4-month and 5-channel broadcast video archive, the total length
of which was 20 months. The algorithm mined CFs from this 20-month stream in
around 5 h.
Figure 1.5 visualizes the chronological distribution of a beer CF. The horizontal
axis indicates the hour of the day, and the vertical one indicates the day of the week.
The darker-shaded blocks indicate the time zones with a higher broadcast frequency.
A well-regulated distribution can be observed from this graph. The distributions
with the same figuration were also found in the graphs of other alcohol-related CFs.
This is because there is a voluntary restriction in Japan, which prohibits the broad-
cast of alcohol CFs from 5 a.m. to 5 p.m. on weekdays and from 5 a.m. to 11 a.m. on
weekends.
Figure 1.6 shows an example of a car CF. Three time zones with higher broadcast
frequencies can be seen in Figure 1.6a. The first time zone is from 6 a.m. to 8 a.m.,
which is when most salaried employees are having breakfast. The second one is
around 12 a.m., which is lunch time. The third one is after 6 p.m., which is when
most salaried employees have finished their work and are watching TV. This obser-
vation can also be found in Figure 1.6b. One explanation of this observation could be
that this car targets toward male rather than female drivers. Another explanation
could be that in most Japanese families, the fathers rather than the mothers manage
the money, so the car dealer prefers the most popular TV slots for male viewers
rather than those for female viewers.
By analyzing these statistics derived from CF mining, the actual conditions of
the business management, which could not be efficiently grasped by using conven-
tional technology, can be determined in a business-like atmosphere. Furthermore,
the influences that the information sent from producers, distributors, and advertis-
ers had on the audiences and consumers can be circumstantially analyzed by
a nalyzing the relationship between the statistics from the TV commercials and
those from the Web sources, including blogs, and Twitter. We believe that this
kind of knowledge can be transformed into business intelligence for producers,
distributors, and advertisers.
61
(a) 61
54 53
48 47
Broadcast frequency
42 40
36 33 32
30 29
26 27 28 27
25
24
19
18
13
12 9
6 7
6 5 4 4
0 1 0 0
00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Hour of the day
(b)
Sat 9 7 1 1 3 2 2 7 6 4 6 5 4 4 1 18 2 3 4 5
Fri 6 3 4 10 3 1 10 1 7 9 12 8 13 5
Day of the week
Thu 7 5 8 6 10 12 3 12 7 3 13 2 7
Wed 5 4 6 4 3 6 1 1 5 14 2 2 1 2
Tue 4 4 2 2 4 1 9 3 2 6 5 6 11 3
Mon 1 5 4 4 1 1 5 1 1 6 7 5 7 1 5
Sun 9 2 1 2 1 6 5 6 3 8
00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Hour of the day
Figure 1.6 Daihatsu Tanto Exe CF chronological distribution. (a) Hour–frequency histo-
gram. (b) Hour–day frequency diagram.
1.4 News Story Retrieval and Threading

1.4.1 Background
The spread of digital multimedia content and services in the field of broadcasting
and on the Internet has made data retrieval and its reuse an important methodology
for helping a user to more easily grasp the availability of these sources. A recent
research domain known as news story retrieval has emerged largely in response to
this methodological trend, and it is a fundamental step toward semantic-based video
summarization and browsing. It tracks the development of evolving and historical
news stories originating from a specific news topic, mines their dependencies from
different sources, and represents their temporal flow in a semantic way.
News story retrieval is normally studied as a Query-By-Example (QBE) topic with
textual information as the underlying cues. A QBE parser parses the query, for example,
the text of a full document obtained from the given Web pages or the closed captions of
a news story, and looks for keywords on the basis of which similar documents are
searched for (Chieu and Lee, 2004; Ide et al., 2004; Kumar et al., 2004; Lee et al., 2010).
The main limitation of this research domain is that the textual information is normally
not discriminating enough to distinguish between the documents of similar but irrelevant
topics. For example, a news story on the “Trial of Saddam Hussein” was broadcast on
November 6, 2006 during the Japanese news program “FNN SPEAK.” The keywords
parsed from this story included “sentence” (8), “this go-round” (4), “former president
Hussein” (3), “President Bush” (3), “United States” (3), and “November 5th” (3), with the
parenthetic numbers being the frequencies of the keywords. By using these keywords as
the query and Ide et al.’s work (Ide et al., 2004) as the search engine, the news stories on
irrelevant topics were output as the results, including the “Sentence of Homicidal
Criminal Yasunori Suzuki,” “United States Midterm Election,” “Sentence of Criminal
Miyoko Kawahara,” and “Sentence of Homicidal Criminal Shizue Tamura.” The reason
for this is because the keywords parsed from the query were limited and not informative
enough to represent the characteristics of the corresponding news topic. In addition to
the textual transcripts, news videos provide a rich source of visual information. In
broadcast videos, there are also a number of visual near duplicates, which appear at dif-
ferent times and dates and across various broadcast sources. These near duplicates basi-
cally form pairwise equivalent constraints that are useful for bridging evolving news
stories across time and sources.
The aim of this research is news story retrieval and reranking using visual near-
duplicate constraints. A news story is formally defined as a semantic segment within
a news program that contains a report depicting a specific news topic. A news topic
is a significant event or incident that is occurring in the contemporary time frame.
A news story can be described in the form of multiple video segments, for example,
shots, and closed captions. Each of the segments and sentences in a news story must
be semantically identical with respect to its news topic. For instance, a news program
can typically be regarded as an assembly of individual news stories that are presented
by one or more anchors and collectively cover the entire duration of the program.
Adjacent stories within a news program must be significantly different in topic.
Figure 1.7 shows an example of two news stories reported by different TV stations
and on different days.
The first task of the proposed news story retrieval and reranking approach involves
a news story retrieval process based on the textual information. Given a news

story sq as the query, the objective is to return a set of candidate news stories

S = {si | i ∈ {1, 2, … , ns }}, satisfying the following criterion:

Sim( sq , si ) > σ q ∀i ∈ {1, 2, … , ns } (1.1)
   
Sim( si , s j ) denotes the textual similarity between si and s j , σq a threshold, and ns
the number of candidate stories. The second task involves a news story reranking
process based on the pseudo-relevance feedback, which is the main emphasis of this
approach. The objective of this task is to find a subset S ⊂ S, then to assume that all

s ∈ S are semantically identical to sq in the topic, and finally, to do a query expan-
sion under this assumption to perform a second-round retrieval. The problems that
need to be solved here include: (1) how to automatically find a proper subset S out of
S satisfying the assumption described above, and (2) how to expand the query so that
the relevant stories low-ranked in the initial round can then be retrieved at a higher
rank to improve the overall performance.
... Lewinsky is claiming that during that visit the president told her that she could testify in Mrs Jones’
lawsuit that her visits to him at the White House were to see his secretary, Betty....
1998/02/24 ABC
... Annan said the “qualitative difference” of this agreement from previous ones is that this one has been
negotiated with Saddam Hussein himself, which means that “the leadership has ....
Figure 1.7 Two news stories originating from “Lewinsky Scandal” and “UNSCOM”
news topics. (From LDC, TRECVID 2004 Development Data. With permission.)
1.4.2 StoryRank
We propose a novel approach called StoryRank in this study, which solves the first
problem by exploiting the visual near-duplicate constraints. After retrieving news
stories based on the textual information, near duplicates are mined from the set of
candidate news stories. A one-to-one symmetric algorithm with a local interest point
index structure (Ngo et al., 2006) is used because of its robustness to any variations
in translation and scaling due to video editing and different camerawork. Figure 1.8
shows some examples of visual near duplicates within the set of candidate news
stories depicting the “Lewinsky Scandal.” Note that some of these examples are
derived from news stories reported by different TV stations or on different days.
Visual near-duplicate constraints are integrated with textual constraints so
that the news story reranking approach can guarantee a high-relevance quality for
Figure 1.8 Visual near duplicates existing across multiple news stories for same topic.
(From LDC, TRECVID 2004 Development Data. With permission.)
Sq Chronological order
Program 1 S S S S S
Program 2 S S S S S
Program 3 S S S S S
Program 4 S S S S S
S S S S S
S S Feedback set 2 S
S S S S
S S Feedback set 1 Feedback set 3
S : News story
Figure 1.9 Exploit must-link near-duplicate constraints between news stories.
p seudo-relevance feedback. As shown in Figure 1.9, two news stories are linked
together if they share at least one pair of near duplicates. This pair of stories can be
regarded as having a must-link constraint, which indicates that they discuss the same
topic. News stories are then clustered into groups called feedback sets by applying
transitive closure to these links. One of the characteristics of these feedback sets is
that the visual near-duplicate constraints within each feedback set usually form a
strongly connected network so that the news stories that are identical to the query in
topic comprise a majority of all the stories in this network. Another characteristic is
that the dominant feedback set is usually identical in topic to the query. Therefore, the
dominant feedback set is chosen as the pseudo-relevance feedback and used for per-
forming the second-round news story retrieval.
Within graph theory and network analysis, there are various measures for the
centrality of a vertex within a graph that determines the relative importance of a
vertex within the graph, for example, how important a person is within a social net-
work, or in the theory of space syntax, how important a room is within a building or
how well used a road is within an urban network. One of these measures is called the
Eigenvector centrality. It assigns relative scores to all the nodes in a network based
on the principle as follows: connections to high-scoring nodes contribute more, to
the score of the node in question, than equal connections to low-scoring nodes.
PageRank (Brin and Page, 1998) is a variant of the Eigenvector centrality measure,
the most famous application of which is the Google Internet search engine. In this
study, PageRank is used as a pseudo-relevance feedback algorithm and integrated
with visual near-duplicate constraints. Each news story is regarded as a node and the
whole candidate story set a node-connected network. The news stories in the
dominant feedback set are regarded as high-quality nodes, which are different from
the classic PageRank, and for each node in the entire network, only the connections
between the node and the high-quality nodes are used as votes of support to iteratively
define the overall relevancy of the story.
1.4.3 Experimentation Evaluation

The approach was tested by using a very-large-scale broadcast video database
comprised of videos broadcast from April 1, 2005 to March 31, 2009. These videos
were broadcast from eight news programs produced by four Japanese TV stations.
The database was comprised of 100,573 news stories, and 20 queries were selected
for experimentation, as listed in Table 1.1, including 11 Japanese and nine foreign
news stories. The design of these queries is based on the most important topics from
both domestic and international news stories during these 4 years. The duration
within which the search was conducted varied from 1 to 15 months. The average
Table 1.1
Twenty Queries Selected for Experimentation
ID Topic Start Date Duration
T01 7 July 2005 London Bombings July 7, 2005 1 Foreign
T02 Trial of Saddam Hussein October 19, 2005 15 Foreign
T03 Architectural Forgery in Japan November 17, 2005 2 Domestic
T04 Murder of Airi Kinoshita December 1, 2005 1 Domestic
T05 Murder of Yuki Yoshida December 2, 2005 1 Domestic
T06 Fraud Allegations of Livedoor January 16, 2006 2 Domestic
T07 Murder of Goken Yaneyama May 18, 2006 1 Domestic
T08 2006 North Korean Missile Test July 6, 2006 1 Foreign
T09 2006 North Korean Nuclear Test October 4, 2006 1 Foreign
T10 Trial of Saddam Hussein November 5, 2006 2 Foreign
T11 2007 Chuetsu Offshore Earthquake July 16, 2007 1 Domestic
T12 2007 Burmese Anti-Government Protests September 23, 2007 1 Foreign
T13 Tainted Chinese Dumplings January 30, 2008 2 Domestic
T14 Tokitsukaze Stable Hazing Scandal February 7, 2008 1 Domestic
T15 2008 Sichuan Earthquake May 20, 2008 1 Foreign
T16 Akihabara Massacre June 8, 2008 1 Domestic
T17 Wakanoho Toshinori August 18, 2008 1 Domestic
T18 2008 Chinese Milk Scandal September 22, 2008 1 Foreign
T19 Murder of Former MHLW Minister November 18, 2008 1 Domestic
T20 2008 Mumbai Attacks November 27, 2008 1 Foreign
precision (Turpin and Scholer, 2006) was computed to evaluate the performance of
the news story reranking.
The Mean Average Precision of the news story reranking approach was 96.9%,
which was higher than all the related studies based on textual information. An
intuitive reason for this is that the reranking approach based on PageRank improves
the informativeness and representativeness of the original query, and the visual near-
duplicate constraints guarantee a higher relevance quality than the textual constraints
used in pseudo-relevance feedback algorithms. Additionally, evolving news stories
usually repeatedly use the same representative videos even if the airdates and
consequently the subtopics of these stories are distant from each other. Therefore, the
feedback set formulated from visual near duplicates can guarantee a larger coverage
of subtopics than those derived from textual constraints, so the news reranking
approach is more robust against the potential variation of the subtopics of the
evolving news stories.
1.5 Indexing for Scaling Up Video Mining

1.5.1 Background
A major and common issue of works focused on videos and images is to deal with
some very large amounts of data. Millions of hours of video and billions of images
are now available on the Internet and in the archives. Various research domains
share the necessity of finding some similarities in datasets in order to highlight the
links for categorization, classification, and so on. Performing a rough similarity
comparison is a basic quadratic operation. Consequently, applying such a brute force
approach on a very large pool may become hopeless. Some subsets can and must be
predefined, but, even in very specialized domains, the accumulation of data and
knowledge pushes the time consumption of algorithms beyond an acceptable limit.
On the other hand, parallelism can also be performed, but this usually induces a
massive expenditure of money and energy. Moreover, it is a linear solution, and some
other bottlenecks may appear. One solution more than the others has been found to
quickly perform a similarity search. The core of the proposition is to perform an
approximative similarity search, which, for a low loss of quality in the results, may
offer a tremendous reduction of time consumption, which is better than for an exact
similarity search. A large part of these approximative approaches use an index
structure (Datar et al., 2004; Joly et al., 2007; Lv et al., 2007). This type of structure
is used to filter out the database, and thus shrinks the pool of candidates.
Different approaches must be investigated depending on resources and requirements
in order to organize a video database. The similarities between videos can be computed
between their annotations, their audio component, or their image component.
A textual search engine can be set up if some annotations are available. The main
advantage of such an approach is the powerful scalability. It can also deal with the
semantic aspect provided by the users. The drawbacks are the sparsity and low
resolution of the annotations. The audio part of a video can also be processed (Haitsma
and Kalker, 2002); lots of speech-to-text applications are gaining popularity, and a
more accurate extraction of the meaning and knowledge is being attained. Videos
sharing the same sound line, or even subject, but different visual contents can be
gathered. However, when the sound line is changed between a copy and an original, it
becomes harder to gather similar ones. In this case, the image part is processed (Ke
et al., 2004; Joly et al., 2007; Law-To et al., 2007; Poullot et al., 2008). The advantages
and drawbacks of audio and image approaches are somehow opposite to the textual
ones: a potential tight resolution, but a lack of scalability. The following part of this
section is dedicated to one of the possible image-based solutions.
In this proposition, the goal is to find videos with a very strong similarity: copies,
that is, near duplicates. Digging out copies can be useful in many automatic processes
for organizing and preserving the contents: merge annotations, clustering, cleaning
(by deleting some duplicates), and classifying (using nodes and links attributes).
Moreover, once links are found, some mining works can be done to characterize
them and the videos. Now, given the massive amount of video available contents, a
scalable approach is mandatory for finding the links between videos.
A cross-dimensional index structure (Poullot et al., 2008, 2009) based on the
local description and image plane is introduced here. It is designed for images but
can be used for video (an image stream) by extension. The structure can easily be
generalized to many applications; here, it is applied to a content-based video copy
detection framework.
1.5.2 Glocal Description
The Glocal description relies on a binary quantization of the local description space.
This space is considered a cell, and each step of the quantization splits the cell(s) of
the previous step along one dimension. The number of cell is Nc = 2h, where h is the
quantization step. Figure 1.10 shows three possible quantizations of a description
space (symbolized in 2D here, and should be 128D for a SIFT case) containing six
local descriptors. The resulting Glocal vector is binary, the presence or absence of
descriptors in the cells is quoted. It is not a histogram, so it is not influenced by the
textures and repeated patterns, which is the inverse to standard BoW features.
The Dice coefficient is used to measure the similarity between Glocal signatures,
SDice (g1, g2) = 2 × (G1 ∩ G 2)/(|G1| + |G 2|), where Gi is a set of bits set to one in the
signature gi and |Gi| denotes the set cardinality.
There must be a balance between the number of local features used and the
quantization in order to obtain discriminant vectors. Therefore, sparsity is needed,
for example, in Figure 1.10, Nc = 16 at minimum is needed. For the case of a copy
1 3 9 11
1 3 1 3 5 7 2 4 10 12
5 7 13 15
2 4 2 4 6 8 6 8 14 16
1111 11000111 100010000100101
Figure 1.10 Three different quantizations of six local descriptors at Nc = 4, 8, 16.

detection, using Na ∈ [100, 200] local descriptors per keyframe and Nc = 1024 has
shown good results. However, for an object search application, a lot more local
descriptors must be picked up in the frames, hundreds (or more) of them, and thus Nc
must be increased to more than a thousand in order to preserve the sparsity and
discriminating power of the vectors.
Two types of descriptors were used in our experiments, a local jet spatio-temporal
descriptor, which is a 20-dimensional vector (see Gouet and Boujemaa, 2001 for
details) and SIFT. The first one is computed at the Point of Interest (PoI) detected by
the Harris Corner Detector. SIFT is computed at PoI detected by using the Difference
of Gaussian. However, using raw SIFTs for building the Glocal descriptors is not a
good option, because of the sparsity of the SIFT description space itself. So, the
SIFTs are first randomly projected in a subspace of 32 dimensions that maintains the
intrinsic dimensionality of the SIFTs.
Then, for the rough local jet and projected SIFT descriptors, the components are
ordered by decreasing uniformity (computed on a set of random descriptors). Indeed,
as only the 10th or so first dimensions are used for the quantization, it is better to
work on the ones that offer the best balance and separability.
Only the keyframes are exploited to create the descriptors to improve the
scalability of the system. They are automatically detected in the videos using the
global luminance variation, and the appearance rate is 3000 per hour (a little <1
keyframe per second, very much less than the usual 25-fps rate).
1.5.3 Cross-Dimensional Indexing
Once the Glocal descriptors have been computed for the entire video set, the indexing
is performed. The mining task has a quadratic complexity (Θ(n2) where n is the num-
ber of items). Only the keyframes and a Glocal description are used to reduce n and
the complexity of the task. Indexing aims at further reducing the final complexity.
The idea is to construct buckets based on the N-grams of a bit set to one in the binary
descriptor. This scheme was already tested before by picking the bits set to one
depending on their position in the binary vector itself. For instance, take the positions
of the three following bits set to one to build a triplet. This triplet is associated to the
bucket where the Glocal descriptor is inserted. In that way, the triplets 1-6-11, 6-11-14,
and 11-14-16 from Figure 1.10 can be obtained, and four buckets in which the Glocal
descriptor of the keyframe is inserted.
In the case of strong occlusions (half picture for example), this generally leads to
a loss of detection. That is not surprising as the local descriptors are not associated
considering their position in the keyframe but in the description space for indexing.
A dual space indexing which uses both the position in the keyframe and in the
description space is then more likely to be relevant, that is, stronger to many postpro-
duction transformations observed on copies, as occlusion, crop, subtitles, inlays, and
so on. The recall gets better, and, crossing two types of information for indexing
precision does as well. Note also that in the case of an object search, this solution will
also be much more relevant, and will be able to focus on only a small part of the image.
The local descriptors in a keyframe may describe the background, a moving
object, a standing character, and so on. Linking the local descriptors to objects is
very difficult, and thus, the community for this issue has not yet been determined. It
would require perfect segmentation and good tracking. Moreover, these tasks are
likely to have higher computational costs, and so, are not suitable for the scalability
issue. However, using these relations between the local descriptors would be very
powerful. In order to exploit this relation, a simple assumption can be made: if the
positions of the PoI are close to each other the local descriptors computed around
are more likely to describe the same object in the keyframe. Consequently, descrip-
tors that are close must be associated, and we propose using these associations in the
indexing scheme. The quantizations of the local features (the bits set to one in
the binary vector) are thus associated depending on the positions of their PoI in the
keyframe for building the N-grams. The indexing relies on a 2-fold association that
crosses the description and keyframe spaces.
Let say N = 3, the 3-grams are built this way: take a PoI, find its nearest neighbor
in the keyframe (using L2 distance in the image plane), then associate it with 1-NN
and 2-NN, and with 3-NN and 4-NN, and so on. The resulting 3-grams are the quan-
tizations of these local descriptors associations. Of course N can be increased, but
the higher it is, the stronger the constraints are: N PoI neighbors must be preserved
and the descriptors computed at these positions must be quantized in the same way.
As PoI or SIFT matching can be lost between two versions of a video, creating
only one N-gram per local descriptor is still quite risky and unstable for retrieval
purposes. Some more redundancy can be injected in the index for improving the
recall. For each descriptor, an unlimited number of N-grams can be set up using
more or less neighbors. The more N-grams that are chosen, the better the recall is
and the lower the scalability is. In Figure 1.11, on the left is a set of PoI, and three
N-grams from two of them, and on the right the N-grams resulting from the deletion
Figure 1.11 Set of points with three triplet associations, centered on two different PoI. The
original associations are on the left, and the deletions and insertions of one point are on the right.
Exploring the Variety of Random
Documents with Different Content
culture—that balance which might perhaps have produced the most
brilliant civilisation in history—will never be perfectly secured, at least
so long as the conditions of the world remain what they now are.
America will be able to continue her Europeanisation, and Europe her
Americanisation, as we have seen is the case. But this interchange of
influences will not have as its only result the increase of the wealth of
Europe and of the culture of America. It will give birth in the two
worlds to a discontent and an unrest which nothing will be able to allay.
In fact, the more its upper classes come under the influence of the
culture of the Old World, the more ardent will become the admiration
and desire of America for the antique,—for that beauty which the
civilisations preceding our own created with such wealth and perfection
of forms; the more strongly will it be convinced of our artistic
inferiority, and persuaded that one of the treasures of life we have lost
irreparably, and can only enjoy in the relics of other generations. Every
picture or statue or artistic object which crosses the ocean and enters
America, every museum which a rich Maecenas opens in the New
World to the public or which a city or a state creates, every chair
founded, and every book of artistic history printed,—everything, in fact,
which brings the spirit of America into contact with the masterpieces of
ancient European art, and awakens recognition and admiration for
them, makes America at the same time recognise how comparatively
decadent is that which our times produce; reveals to it that lost
Paradise of Beauty around whose closed gates we are now condemned
to hover. This contact with the artistic achievement of the past
becomes, therefore, at the same time a gadfly of discontent and unrest
to the upper and cultured classes.
By an inverse process, the further the spirit of American progress
penetrates into Europe, the more completely does it detach the Old
World from its past, and, therefore, irritates, grieves, and disgusts the
classes which have enough culture to recognise and admire the
marvellous arts of that past. With every ten years that elapse, we feel
that our past, with all its radiant glories, has receded a hundred years;
that we are plunging into a new world, in which riches, knowledge, and
our power over nature will increase, but which will be ugly,
inharmonious, and vulgar compared with the centuries which preceded
it. Many Americans fail to understand why Europe cherishes so many
latent antipathies to America, which has never done her any harm,
directly at least. The real reason for these antipathies must be looked
for in the artistic decadence which accompanies the development of
modern civilisation. That civilisation has not been created by America
alone, but by America and Europe combined. Europe and America,
therefore, share the responsibility for this decadence. Yet it suits the
European book from time to time to see in America the symbol of the
civilisation of railways, steam, electricity, business, and industry on a
large scale; and Europeans gladly vent on America their spleen for
whatever in this civilisation, pregnant with good and with evil, offends
them and arouses regret.
In short, there is an insoluble contradiction between progress, as
our age understands the word,—between the “American” progress, as
many Europeans call it,—and art. This contradiction has as yet
attracted but little notice, in the still great confusion in which we live, in
the initial tumult of this new civilisation which is invading the earth. It
will be noticed, however, more and more strongly, as generation
succeeds generation, and in many families the primary hunger for
wealth is satiated and gives way to the desire to “translate quantity into
quality”: as American love for the historical beauties of the Old World
increases, and Europe takes further lessons in the multiplication of her
wealth. There is no escape or salvation for our civilisation from the
discontent and unrest which will arise from this contradiction. It is a
torment which will grow with the growth of wealth and culture; which
nations and classes will feel more acutely the richer and the more
cultivated they become; from which perhaps, one day, America will
suffer more severely than many European nations; which will oppress
the upper classes much more heavily than the people. The latter,
indeed, will not feel it at all, and will alone be able to live in modern
civilisation, as contented as man ever can be in this world.
History often has strange surprises in store. The civilisation of
machinery tended at its birth to appear as a death-blow to the working
classes, a godsend to the upper classes. For years and years, socialism,
generalising from the initial rubs, predicted and pretended to prove that
the great mechanical industry must enrich a small oligarchy
inordinately, and reduce to the blackest wretchedness the great mass
of the population; that a new feudalism of capitalists, fiercer than the
barons of the Middle Ages, would seize all the good things of the world.
A century passes, and we find this civilisation giving complete
satisfaction only to the workmen, because it can content the workmen
only from the double point of view of quantity and quality. It gives
them an abundance which only a small fraction of the people enjoyed
up to a century ago; and, at the same time, bestows on them a luxury
which fully satisfies their yet simple and unsophisticated æsthetic
sense. We may smile when we see in a workman’s home mirrors and
clocks which are the rudest imitations of masterpieces of the Louis XV
and Louis XVI styles, hideous reproductions from a German factory;
and coarse carpets which are poor European copies of beautiful Turkish
and Persian models, the result of the substitution of the iron teeth of a
machine for the industrious fingers of the human hand, and of
decadent aniline dyes for the brilliant and unfading vegetable colours.
The workman, however, does not know the matchless models of which
these objects are the ugly copies, and, inasmuch as every æsthetic
verdict arises out of a comparison, these reproductions represent for
him the summit of perfection, and entirely satisfy his need for beautiful
things round about him.
To the upper classes, on the other hand, this civilisation has given
immense and imposing wealth, such as no epoch had ever considered
possible; but it has deprived them of the means of enjoying it. Wealth
becomes nowadays more useless, the greater it becomes, because a
multi-millionaire cannot build himself a house, wear clothes, or buy
objects a hundred times finer and better than the possessor of only a
few millions. We no longer have artists capable of accomplishing
miracles. The men of great wealth are forced to compete with each
other for the relics of past beauty at fabulous prices, when they are not
inclined to spend all their wealth for the benefit of others. These relics
are not sufficient, however, to satisfy the desire for beauty and art
which grows in our times with the growth of wealth and of culture. On
the contrary, they only make us feel more acutely the decadent
vulgarity of everything, with but few exceptions, which our age
produces.
* * * * *
Someone will say that, after all, this torment is not a very serious
one; and that men will easily find means of consoling themselves. No
epoch, as I have already said, can have everything; and modern
civilisation bestows on the wealthy classes of our times numberless
compensations for the ugliness of the modern world. One of these
should be enough by itself to content even the most discontented: that
kind of bodily and mental ubiquity which is enjoyed, thanks to the
prodigious inventions of modern genius. Cannot the wealthy, thanks to
their riches, remove from one continent to another, travel, have
dealings and acquaintances in, and receive communications from, every
part of the world, come to know the most distant and recondite
beauties of nature and of art—in a word, live over the whole globe? A
modern great man may well feel himself almost a demigod compared
with the men of two centuries ago, so great is the sway his money
gives him over the forces of nature, so easily can he escape from the
tyranny which space and time used to exercise over men up to a
century ago. May not the intoxication of this proud sway be worth as
much as the pleasure which the works of Phidias, Michelangelo,
Raphael, and Houdon gave to our ancestors?
It is true that the pride of our knowledge and the intoxication of
our power stun us, and therefore help us to bear with greater patience
the want of more æsthetic pleasures. But this compensation, like all
compensations, is of its very nature provisional. It is not possible
utterly to destroy a need inherent in human nature. There is at the
present day a certain tendency in the world to consider art as a
superfluous frivolity, as a luxury only to be thought of in moments of
leisure. Art and such-like superfluities are contrasted with what are
called the serious occupations, the practical realities of life: industry,
commerce, inventions, business, and wealth. Those who hold this
opinion forget, however, that the sculptures of Phidias and the
paintings of Raphael appeared in the world long before the steam-
engine and the Voltaic pile. Are we, too, prepared in face of this fact to
affirm, with the most advanced champion of the American idea of
progress in my dialogue, that “history was off the track up to the
discovery of America”? That if men had had any sense, they would
have invented machinery and developed the sciences first, and then
created and developed the arts? But even if we were prepared to
maintain this paradoxical theory, another fact of common observation
would be there to prove to us that beauty is not a luxury and whim of
gentlemen of leisure, but a primary, universal, and indestructible need
of our minds, which every human creature seeks to satisfy as best it
can. Do we not see every day that the peasant and the artisan, items
in modern civilisation though they are, no sooner have a little money
than they try to procure for themselves ornaments, either for their
persons or for their homes, which may be as tawdry as you like, but
which to them seem beautiful and well worth the expenditure of some
of the little money they have? Have we not seen that one of the merits
of modern civilisation is that of satisfying the æsthetic needs of the
masses? Why should we not expect, then, to find the same need,
though in a refined and intense form, felt by those to whom superior
intelligence and energy, or the favour of fortune, has granted the
power to accumulate wealth in large quantities, those, in other words,
who have at their disposal greater means of procuring for themselves
the pleasures and good things of life?
No: the artistic impotence of modern civilisation is likely to prove,
to judge by the first effects which are now beginning to manifest
themselves, a graver phenomenon than is at present realised. The
upper classes in Europe and America will not be able to go on for an
indefinite length of time living with a consciousness that the world in
which they find themselves is ugly, coarse, and decadent in comparison
with preceding civilisations; feeling the inferiority of the present more
acutely the more they study, and at the same time extending still
further their sway over the world and accumulating new riches to
console themselves for it. This would be a state of moral want of
balance; and moral want of balance cannot continue indefinitely, just as
physical want of balance cannot continue indefinitely. Either our
civilisation will abate its aspirations to the level of the mediocrity which
it is capable of producing in art, destroying in itself the remembrance of
and regret for those ancient civilisations which created so many
beautiful things; or it will have to put itself into a position to satisfy not
only the æsthetic needs of the masses, but also those of the more
cultured and refined strata of society. The first supposition appears
improbable, or at least no man of sense will wish to consider it
possible. It would mean a relapse of the world into barbarism, the end
of all the traditions and all the studies which have been and still are an
indispensable element of intellectual and moral refinement. So we are
left with the other hypothesis, which assumes that man will make up
his mind one day to make an effort to create arts of his own, which will
survive comparison with those of the past.
Yet the task is an arduous one. As I have already said, an art is not
created or perfected without the spirit of tradition and of discipline; and
the attempt in our times to re-infuse vigour into the spirit of tradition
and of discipline, if only in the measure necessary for the progress of
art, is an enterprise, the difficulties of which everyone can understand
without a long explanation. It cannot be effected without a profound
intellectual and moral reform, which will bring about a change in many
things besides the originality and power of the arts which are to-day
languid and decadent. So the struggle between American progress and
art may well be a more important phenomenon than is at present
apparent; and may well entail transformations of far-reaching extent.
V
BEYOND EVERY LIMIT
F OR century after century, our civilisation lay low in its

Mediterranean lair. It knew but a small part of Europe, Asia, and
Africa; and no pricks of curiosity impelled it to ascertain how far the
world extended over land and sea beyond the vague bounds which
marked the limit of its efforts. That small part of the world satisfied the
ambitions of our ancestors, though they certainly were not shy and
craven. Confined in that narrow corner of the earth and with only the
scanty resources which it could provide, they created literatures, arts,
philosophies, states, laws, and religions. Some of these creations are
alive to this day, and help us distant descendants to shed a little beauty
and to impose a little order on the modern world.
Between the fifteenth and the sixteenth centuries there began a

great change in the history of our civilisation. Impelled by the wish to
reach India by way of the Atlantic, our forefathers began to explore the
earth. Gradually geographical exploration became the preoccupation of
governments, the passion of the public, and the business of a great
number of persons who made almost a vocation of these daring
voyages. And behold one fine day, one grand day, the most fortunate
and daring of the navigators who were exploring the Atlantic in every
direction discovered America. In mid-Atlantic there lay two connected
continents, stretching from one hemisphere to the other, covering many
latitudes and a great variety of clime, and much of this vast territory
was still but sparsely inhabited. It was then that our forefathers
realised how vast and rich was the earth, and how small and poor in
comparison seemed that Mediterranean world in which for so many
centuries they had lived. It was then that they began to pass beyond
the limits within which they had been so long confined, to invade and
to conquer the outside world.
The Pillars of Hercules, which had been the impassable
geographical limit of the ancient Mediterranean world, were not,
however, the only bounds which they passed; they transcended also
the moral and intellectual limits which until then had circumscribed
their thoughts and actions. All the time that the ancient Mediterranean
civilisation, turned loose into the Atlantic, was striking root and
expanding in America, an uninterrupted sequence of events and
movements was taking place in Europe, destroying ancient laws,
ancient traditions, and ancient discipline; upsetting, in other words, the
bounds placed in the past to the thought, the sentiment, and the will of
Man. The most important of these events were: the Protestant
Reformation, the philosophies of the seventeenth and eighteenth
centuries, the advances of science, the French Revolution and its wars,
the birth and the development of the great industrial movement, and
the all but universal triumph of democratic ideas.
Little by little, while the aspect of the globe changed, a great
revolution was taking place in the spirit of the ancient Christian
civilisation in Europe, which from being dictatorial and traditionalist
became free and progressive. Religion, which for so many centuries
had been a kind of severe moral discipline, a life of prohibitions,
scruples, rules, precepts, ceremonies, and rites, changed into a kind of
free contemplation of the Deity in which the individual conscience had
full play. Everybody became his own high priest.
The ceremonial of social life, which at no time had been so
complicated, serious, and exacting, gradually became so far simplified
as no longer to encumber man in his every movement and activity. The
State, which at one time, hand-in-hand with religion, watched over the
customs and life of its citizens, accorded greater and greater liberty to
them. To-day everybody, provided he contributes to society his daily
sum of work, is free to live and think as he likes. Severe laws used
once upon a time to regulate men’s luxuries and pleasures. Every class
was forbidden to spend its money in any way other than that
prescribed by these laws. There were times of the year in which men
were forbidden to amuse themselves, and, when amusements were
allowed, the laws took care that they should not degenerate into vice.
Nowadays the whole year is one long festival and carnival for all who
have money to spend; and together with freedom in pleasure men
have acquired also a freedom in vice which would scandalise our
friends the ancients, if they might come back to life again.
Every authority is losing power. The people discuss the government
and the laws; children take the first opportunity of escaping from the
authority of their parents. The younger generation is convinced that it
knows more than the older, and values the latter’s experience at zero.
Traditions are losing their force and academics their prestige. Everyone
holds the opinion he likes in religious, artistic, political, and moral
questions; just as he is free to regulate his own conduct, at his own
risk and peril, as he pleases, with the sole obligation of respecting the
limits imposed by the laws, which are for the most part neither
numerous nor embarrassing.
What is the deep-lying cause of this duplex and contemporaneous
movement? Why has the old Christian civilisation of Europe felt itself in
the last four centuries unable any longer either to contain itself within
the ancient material limits or within the ancient ideal limits? Why, at the
same moment as it advances to the conquest of new continents, does
it destroy within itself all the ancient disciplinary restrictions? Because
in these last centuries it has gradually discovered that the earth is
much vaster and richer than it suspected; that it contains, in old
Europe, as well as in young America, treasures in much greater
abundance than it had ever pictured in its dreams; and that it can
invent tools which bring these rapidly within its reach. Gradually, as
man found that he could rob nature of her immense treasures, there
arose and spread from generation to generation through all classes in
Europe and America a craze for wealth and a mad ambition to win the
mastery over nature, such as the world had never yet seen.
To satisfy this craze and this ambition, however, it was necessary to
break many of the innumerable bonds—religious, moral, æsthetic, and
political,—which limited the energy and initiative of our forefathers.
How could so many millions of men have brought themselves to
emigrate to America, if the spirit of tradition had not been weakened in
Europe, and if everybody had continued to hold, as they did once upon
a time, that the greatest good fortune a man could have was that of
being buried in the church in which he had been baptised? Even to-day
there are those who lament the diminution, in all the Christian churches
during the last few centuries, of the number, complexity, and rigour of
the rites and ceremonies, just as others lament that the ceremonial of
social life and etiquette is dying out. But would not men who are
obliged to work, travel, and rush about as we do nowadays find
themselves embarrassed beyond the point of endurance by a religion
which made them spend too much time over rites, and by a
complicated etiquette like that which still prevails in the states of the
East, requiring a large part of the day to be spent in compliments and
ceremonies? Europeans often laugh at the architecture of New York;
and I must confess that I, too, found it distinctly bizarre. On the other
hand, could that vast city have grown and renovated itself so rapidly in
the last century, and have found accommodation for the countless
multitudes which throng to it from all the corners of the world, if those
who built it had troubled themselves to observe the rules formulated by
the great architects of the sixteenth century, in the days when it took
as much time to build a palace and a church as it now does to
construct a city?
Modern society, if compared with the societies which preceded it,
may seem in many of its aspects—and in fact it is—ugly, poor in artistic
beauty, coarse, and brutal. It may even seem atheistic and irreverent,
frivolous and superficial in matters of religion, and, in certain respects,
morally lax or downright licentious. This kind of disorder, however,
which is such a common subject of heart-burning, is only the necessary
effect of the outburst of our energy over the world and nature on its
path of conquest. A civilisation cannot produce, refine, or perfect arts
or traditions of elegance and of social life, or a morality and a religion,
if it does not adopt an attitude of reserve, if it does not limit itself to
some extent, if it does not sacrifice its other ambitions and aspirations
to this object. A civilisation like ours, whose supreme aspiration it is to
extend in the shortest possible time and as far as it can its empire over
the world; to surpass all the limits which nature seeks to oppose to its
restless ambitions and to the multiform energy of man, must needs
sacrifice beauty, refinement, elegance, and moral delicacy to rapidity,
energy, activity, and daring. The discovery and development of new
countries, the marvellous progress of America, the discoveries of
science, the perfection of machinery, the ideas of liberty which emerge
triumphant from political revolutions and changes in customs, the
weakening of the spirit of authority in every department of social life,
the abolition of so many limits which once entangled the movements of
man, are all phenomena which are mutually and indissolubly
connected.
* * * * *
“Yes,” many will say; “and they are phenomena which all go to
make up the grandeur and glory of the modern world. We have power,
wealth, knowledge, and liberty, the four blessings of which our
forefathers had little, if any, knowledge. What cause have we then to
grumble? And amongst all the blessings which modern times shower
upon us, perhaps the most precious—more precious than wealth,
power, and knowledge—is liberty. If we have no reason to regret the
past, it is chiefly because our forefathers lived, imprisoned, and in
suffering, within limits which we have overstepped. Is there any greater
joy for a man than that of being able in thought, feeling, and action to
follow the inner impulse of his own conscience, instead of making it
bend to an external will, whether it be that of the law, or that of the
public, or that of a tradition? Surely the modern world is the greatest as
well as the most fortunate which has ever existed?”
That is what many people think, and thoughts like this breed the
optimism which at the present day cheers so many minds. This thought
is, moreover, partially true; but only partially. For in the intoxication of
their triumph over nature, of the riches which they have conquered so
easily and in such abundance, men seem not to recognise that this
civilisation without limits is little by little allowing that same unbridled
energy to hurry it into excesses which threaten to drive it back into that
very state of barbarism from which it has made so many efforts to
escape. The impetus which it has acquired, now that it has cast off so
many of its ancient restraints, is great; but the danger is precisely that
this impetus may carry it too far.
I have already said that amongst the limits abolished by modern
civilisation are those which preceding civilisations had placed on luxury.
How great a change has taken place in men’s ideas on this subject
during two centuries! Simplicity and austerity were considered for
centuries virtues proper to saints and heroes. Christianity had gone so
far as to glorify poverty in so many words. Man, by increasing his
needs, only increased the number of his masters and tyrants; only
multiplied for himself occasions for sorrow. The more simply a man
could live, the freer, stronger, and happier he was. In short, in ancient
times, up to the French Revolution, religion, law, and tradition set limits
on every side to man’s desire to possess and to enjoy; and these limits
were so numerous and so close, that they entailed no little suffering on
the generations constrained to live within them. That is why we have
upset them all. And what is the result? That we no longer have any
sure criterion by which to distinguish reasonable consumption from
insensate waste, legitimate need from vice. We can no longer say what
are the limits at which it is reasonable and wise for the peasant, the
artisan, the small tradesman, the man of leisure, the millionaire, the
multi-millionaire, the child, the woman, the old man, respectively, to cry
a halt to their desires. All men and all classes arrogate to themselves
the right to desire, to spend, even to waste as much as they can. No
one has any clear idea of a standard by which to distinguish what he
may desire and what he ought to deny himself. A kind of universal
prodigality is becoming obligatory in every class; and modern
civilisation is hurrying towards an unbridled, gross, and oppressive
orgy. There is already a large number of men in Europe and America
who eat, drink, and smoke to excess; who over-indulge in intoxicating
and stimulating drinks; and who spend themselves in that continual
whirl of diversions and distractions which form so large a part of
modern life. The number is fated, however, to grow yet larger, rapidly
and indefinitely. Is not production increasing on every side? Is not
progress for us first and foremost the continual increase of production?
And what avails it to produce more, if the riches produced do not find
consumers?
The modern world, in freeing men’s desires from all the ancient
limits and restraints, has given a vigorous impulse to human industry.
In order to satisfy the increased needs of the masses, man has
invented machinery, and has put a premium on the new countries. But
precisely because there is no longer limit or restraint to men’s desires,
industry, which in the past was the handmaiden of human needs, is
now becoming their tyrant. It is creating and multiplying our needs
with a view to their subsequent satisfaction. In order that it may never
be short of work, it is tempting men in a thousand ways to desire and
to consume more. Therefore our civilisation has made of riches not the
fitting means of satisfying reasonable and legitimate needs, but an end
in themselves. We are obliged to produce them in order to consume
them, and to consume them in order to produce them. Every moment
which a man does not spend in producing riches, he must spend in
consuming the riches produced by others; so that he can never stay
still for one instant, but must jump from occupation to amusement, and
from amusement back again to occupation. He must try to make the
day as long as possible, accustoming himself to do everything at full
speed, and cutting down the hours of sleep as much as possible.
Everybody knows that we moderns, especially in the great cities, are
losing the habit of sleeping.
We have not yet mentioned, however, the most serious drawbacks
of the present-day lack of any fixed limit to men’s desires. In the past
ages, the efforts of religion were directed to educating men to self-
introspection; to teaching them to explore their own consciences, to
render account to themselves of their own sins and vices, and to try to
amend them. One might even go so far as to say that from one point
of view Christianity was principally a melancholy meditation on the
perversity of human nature, and an effort to purify it through
meditation, suffering, and the love of God. One has only to read the
letters of Saint Catherine, or the Divine Comedy, or Pascal’s Pensées to
realise to what an extent the moral refinement which is the fruit of
these meditations preoccupied the loftiest minds, and, at second hand,
the great ones of the earth in past centuries. A considerable part of the
energies of every generation was consumed in this introspective effort,
instead of in action; for centuries and centuries, saints, moralists, and
preachers abounded in Europe, while men of action, fit to conquer the
world and its riches, nature and her secrets, were scarce.
This searching of the inward parts was not always soothing by any
manner of means. For the past century and a half, numerous writers
and philosophers have denounced it as one of the refinements of
torture with which religion in the past made men’s lives a burden to
them. Perhaps, however, they are wrong; for this effort, which religion
made for so many centuries to habituate man to self-introspection, self-
knowledge, and self-judgment, demands a less superficial explanation.
However great be the force of the laws and the vigilance of public
opinion, there can be no convenient order in a social system, if man
does not help by exercising some sort of surveillance over himself, if he
does not give ear to an inward voice, forbidding him to take advantage
of every opportunity of doing evil with impunity which may offer itself.
This necessity for self-restraint is particularly urgent in connection with
three duties: the duty of speaking the truth, the duty of checking one’s
own inclination to pleasure, especially in the relations between the
sexes, and the duty of not using one’s own strength improperly at the
expense of the weak. Many are the times when we could tell a lie with
impunity, or even with advantage to ourselves, if we wished to do so;
and yet it is necessary that we should speak the truth spontaneously, in
order that justice may triumph. How easy it is for the man who has
become the slave of vice to evade the eyes of his fellows and to satisfy
in secret his most perverse passions! And what system of laws can be
conceived which will be wise and perfect enough to bar all the
countless ways in which the stronger can impose upon the weaker?
Every religion with more or less success—and none with more
success than Christianity—in centuries past helped law and public
opinion to regulate this most important part of morality. They all made
a sacred thing of an oath, which is nothing but a covenant which every
individual enters into with himself to speak the truth, even when he
could lie with impunity or with advantage to himself. They all created a
sexual morality to regulate love, marriage, and the family. They
endeavoured in various ways to awaken in the consciences of the rich
and powerful the recognition of certain duties of moderation and
charity towards the weak and the poor. Nowadays, on the contrary,
men no longer have time to examine their consciences, or to reflect on
their own vices and defects, or on their own duties and rights. The
whole atmosphere of our lives is exterior to ourselves; we are always
moving about and always busy. We have become almost incapable of
meditation and self-introspection. Our times no longer lay any store by
this education of our inner feelings. The only discipline they impose on
man is that of work. Everybody, whether of high degree or of low, is
required, under penalty of losing his daily bread or of dropping in the
social scale, to fill with exactness, precision, diligence, and correctness,
the rôle, be it little or big, which has been allotted to him in the
immense operations of our times. But, for the rest, everyone is to-day
much more free than he was in the past to adjust his line of action to
his own beliefs, and to make for himself his own standard and his own
laws.
The result is that all the scruples and internal restraints with which
religion endowed the conscience of man in the past are growing rusty
from disuse. Our civilisation, rich and splendid as it is, threatens to be
spoiled by fraud, by evil habits, and by oppression. There is no doubt
about it; not even in these days is the discipline of work sufficient in
itself to keep the State in good order. Man is not a living machine,
destined only to produce riches. When he leaves his office and comes
back into the world, the modern man there finds a family, sons,
parents, friends, persons of the other sex who may attract him, men
richer and more powerful than himself, others weaker and poorer,
political institutions and public problems; in short, opportunities of
doing good or evil, temptations dangerous but agreeable, and duties
painful but necessary. And our times not only give him practically no
moral assistance to conquer these temptations and to perform these
duties, but rather in many ways incite him to yield to the temptations,
and to exercise his cunning in evading the duties. Fraud in particular is
becoming simply second nature to our civilisation. What is the great
industrial movement of modern times but a continual deception for
cloaking the deterioration which it is bringing about in the quality of
things as the price of increasing the quantity of them? Every day sees
an increase in the number of cleverly faked objects, which are not what
they seem; and science—especially chemistry—is the highly paid
accomplice which furnishes industry with the means of imposing this
colossal deception on an inexperienced and ingenuous public. In other
words, commerce and industry, which play so large a part in modern
life, are becoming more and more a colossal deception in which he
succeeds best and makes most money who is cleverest at lying to the
public and at foisting on them goods of inferior quality though superior
in outward seeming. Now if we see in a social system, on the one
hand, a weakening of all the internal restraints which keep a man from
lying and cheating, and, on the other, a premium put on that same
lying and cheating, must we not expect to find fraud permeating the
whole system? And what will our customs be like, what will life be like,
in the days when nobody any longer feels any remorse or scruple in
cheating his neighbour, and when everybody becomes cheat and
cheated turn and turn about, cheat in matters which he understands,
cheated in those in which he has to rely on other people?
The growing depravity of customs, furthermore, threatens us with
no less a danger. I do not wish to exaggerate the horrors of the
modern Babylons, as Catholic priests and Protestant ministers are apt
to do. Their grief at seeing the rising generation turn a deaf ear to their
wise counsels makes them take too gloomy a view of the present state
of affairs. Nevertheless, it is certain that the customs of modern
civilisation are hurrying it towards a dangerous crisis. The internal
restraints are being relaxed, and temptations and facilities are
multiplying with the growth of riches and of cities, and with the
increasing mobility of persons of both sexes, so many of whom it
prompts to leave their native village or country. Especially in the big
cities where everyone is unknown, can easily hide away, and is watched
by nobody; where money has greater power over men’s minds because
there is more of it and more of it is needed,—virtue runs serious and
continual risks. Without being aware of it, we are undoing, little by
little, Christianity’s great contribution to the chastening of our customs,
by suppressing many of the limits which Christianity had established
with such labour in the midst of the unbridled licence of the ancient
world. We are travelling, therefore, step by step back towards
paganism, with all its conveniences and all its perils. Already, in fact,
we can see cropping up here and there in the richer and more highly
civilised countries and classes that mortal sickness which killed the
ancient civilisations: sterility. One of the reasons why all the most
flourishing ancient civilisations have perished is that at the moment of
their greatest glory the population suddenly began to dwindle; and this
sterility which killed them was the effect to a large extent of the licence
of their customs. Love remains fertile only so long as it restrains itself
and limits itself. Christianity, by subjecting men’s customs to discipline—
one of the noblest of its services to mankind—succeeded for centuries
in maintaining in Europe and America an incessant fertility, which has
proved to be one of the most potent causes of the increase of our
power. But now we can see, with the return of the world to paganism,
the beginning of a new era of sterility, especially in the big cities and in
the most ancient and most wealthy states.
Lastly, I have referred to another danger which threatens this our
social system, victim as it is of its limitless desires; I mean the increase
in the opportunities for the strong to abuse their strength. This is
certainly the least of the three evils; for thanks to the diffusion of
culture and of liberty, the weak have learned and are able, to unite in
their own defence. Some balance of justice is obtained and will
continue to be obtained by opposing force to force. The balance,
however, will be in external things rather than in men’s convictions. For
in this unbridled and limitless chase after money and enjoyment, of
which the world is the theatre, the spirit of charity is obscured; and
men’s minds become accustomed to a hardness and brutality which
may perhaps one day startle the world in a disagreeable and terrible
way.
* * * * *
It may seem to some of my readers that I take a delight in uttering
gloomy prognostications of the future of modern civilisation. Such,
however, is not my intention. Who would dare to deny that,
notwithstanding its defects, the civilisation in which we have the good
fortune to live is the most splendid and powerful on which the sun has
ever shone? But its very grandeur, which is to so large an extent the
fruit of our boldness in overthrowing most of the limits which preceding
civilisations had placed to human energy, gives birth to a new and
formidable problem which is already beginning to confront our speed-
loving civilisation, and which is itself, too, a problem of limits, perhaps
of the limit par excellence. And that problem may be expressed in one
question: Quousque tandem? Up to what point, in our desire to
conquer the world and its treasures, to multiply riches, and to increase
our power over nature, must we and can we sacrifice beauty, and the
forms, ceremonies, and refinements of life, moral and æsthetic? Up to
what point must we and can we make legitimate use of the liberty
which the modern world has given us; and at what point does abuse of
it begin?
This is the vital problem which I have posed and tried to dissect in
the dialogue which my travels in America inspired me to write: the
problem treated in the speeches and discussions of the many
characters, European and American, who figure in that dialogue. It may
seem strange, at first sight, that a discussion of the Old World and the
New, in which the contending parties propose to prove which is
superior to the other, should end in this second problem, apparently so
unlike the first; whether it is necessary or not to place a limit on the
unbridled activity and immoderate desires of our times. Anyone who
has read the present series of essays, however, will be less likely to find
this conclusion singular and obscure. I have repeatedly said, and tried
to prove, that there is too great a tendency on both sides of the
Atlantic to find an antagonism between Europe and America. If certain
tendencies are stronger in one of the two continents, and weaker in the
other, these are differences of quantity, not of quality. America is
becoming Europeanised, and Europe Americanised. However little
reflection and cool reasoning the European may bring to his abuse of
America on the score of its excessive zeal in the production of riches, or
the American to his abuse of Europe on account of the scanty remains
of the spirit of tradition and conservatism in the Old World, each will
recognise that he is at the same time inveighing against his own
continent. In fact, Europe applies herself with no less zeal than America
to the production of greater wealth; and America is no less anxious
than Europe to enjoy the advantages which may even now accrue to
the world from the spirit of tradition.
Consequently the discussion of the question whether America is
superior to Europe or Europe to America is a futile enterprise and
labour lost; because the balance between the differences is rapidly
adjusting itself. Nevertheless, if any difference exists to-day between
the two continents, it is undoubtedly this: that all the phenomena of
social life in America are simpler and clearer, and less overlaid and
obscured by traditions, institutions, and century-old ideas and
sentiments than in Europe. For this reason, the careful observer will
find in America a much more profitable field for the study of the
dangerous tendencies and exaggerations of modern civilisation which
are common to Europe and America. Of these dangerous tendencies,
the one which has struck me most in the course of my travels in
America, and has given me most food for thought, is precisely this,
which I have treated in this my latest work and which forms, as it
were, the crown to the whole discussion of the dialogue. Modern
civilisation has accomplished miracles and marvels without number,
since she left behind her the limits, material and ideal, within which the
timid generations of old confined themselves,—since she outstepped
and upset these limits on her way to conquer the earth, riches, and
liberty. Now, however, precisely because she has crossed all the limits
and no longer has any before her, she finds herself impelled on every
side, in politics, customs, morals, art, and philosophy, to excesses
which may one day prove very dangerous. Men are beginning to have a
vague presentiment of this danger. They do not clearly see, however,
the quarter from which it threatens. They disquiet themselves without
thoroughly diagnosing the evil. And this disquietude may perhaps
explain the pessimism which afflicts a civilisation so flourishing and
fortunate in many respects as that of our own times.
For this reason, I thought that the great problem of the limits might
grow little by little, on board a transatlantic liner, out of a discussion
about America. An Italian, who has made money in America, and who,
like so many Europeans who have made their fortune thus, is an
admirer of the New World, one evening launches out into a eulogy of
young America at the expense of old Europe. He extols the civilisation
of machinery, progress, and liberty, by contrast with what remains in
Europe of the ancient civilisation whose efforts were directed to
improving the quality of things rather than augmenting their quantity;
which left the world poor while it created arts, religions, moralities, and
rights. The discussion becomes heated, complicated, and diffuse until,
under the guiding influence of an old savant who knows Europe and
America too, it concentrates on this point: Granted that man was well-
advised to exceed the ancient limits within which preceding civilisations
had confined him, to hurl himself on the world and to conquer it; up to
what point may man aspire to liberty in every department of life,
without endangering in the long run the most precious fruits of his
conquest?
The book does not pretend to solve this formidable problem. No
philosopher, no writer, no book could solve it. It can only be solved by a
radical revolution in the ideas, sentiments, and interests of the masses.
But the book which I have written purports to throw light on some, at
least, of the essential aspects of the problem. It endeavours to make it
clear to the men of our time, by harping on a principle of great
antiquity, great simplicity, and great modesty, which may be perhaps
usefully recalled to the memory of present generations, in Europe as
well as in America. That principle is, that man is a being of limits; and
that he ought, therefore, to observe in his desires a certain mean. A
civilisation must remember that it is the sum of the efforts of a great
many individuals; that these individuals may be very numerous, but
that each one is a small limited being; and that the sum of their efforts
cannot be infinite. Consequently, a civilisation must not let its desires
and wishes extend untrammelled in every direction. It must learn to
confine itself within limits.
VI
THE RIDDLE OF AMERICA
I N Argentina, there are vast and luxuriant valleys, over which the
train seems to creep toward the very edge of a horizon which ever
recedes as the traveller advances. From time to time, four or five red
one-storied houses, clustered behind a station, recall to his mind the
fact that this wilderness is actually inhabited. In Brazil, so far as the eye
can see, there are ranges of mountains, shadowy even in brilliant
daylight, in the midst of which, from time to time, one mountain stands
out more distinctly than its fellows. The shadowy hills are those still
covered by the primeval forest; the others, those where the timber has
been burned off and replaced by coffee plantations; but even here
there is no trace of human life. One must travel long hours by railroad
before even catching sight of a village.
In North America, or at least in its Eastern States, there are vast

and desolate tracts. From time to time a village appears, bristling with
chimneys. Then the traveller slips on into the deserted country. Another
village appears, only in its turn to disappear. Then all at once the train
begins to rush through the midst of houses. On, on it goes. The houses
never cease to follow it. Huge edifices rise from the midst of the little
dwellings like giants from a crowd of dwarfs. Automobiles and trolley
cars move through the streets. It is a great city; half a million, a
million, two million men are crowded together there in the shadow of a
thousand chimneys, surrounded on every side by an almost deserted
country. What a strange sight are these wildernesses to a European
accustomed to live in one of the crowded countries of the Old World
where men have built their houses everywhere, from the shores of the
sea up to the highest habitable slopes of the mountains!
In observing a phenomenon so novel to his experience, the
historian of antiquity is deeply interested; and as he studies it, like so
many other Europeans in the presence of the same spectacle, he
forgets his own preoccupations. The riddle of America rises before him
and the desire of finding an answer to it turns him from his former
studies. For America is a true riddle to Europeans. During the past
thirty years, not only the United States, but even smaller American
countries like Brazil and Argentina, have impressed themselves sharply
upon the attention of Europe. The Old World has been compelled to
recognise that America has in her turn become a mighty historic force;
and that she exercises an influence on the Old World which grows
continuously greater. When one reflects that, only a century and a half
ago, all these American states were merely poverty-stricken colonies of
Europe, harshly exploited by their European masters, one cannot
suppress amazement at the rapidity with which their destiny has
changed.
What power is it which has worked this miracle? On this point, it is
impossible to feel any doubt; the power is wealth. These plains and
these mountains which look so deserted are tilled, mined, worked with
intensest energy; and every year, with a generosity which seems
inexhaustible, they yield to the men who have toiled over them
prodigious quantities of cereals, tobacco, coffee, wool, gold, silver, iron,
oil—an enormous torrent of riches which pours over the entire world.
The great industrial cities of North America manufacture these raw
materials with profits so large and swiftly won that to the Old World
they seem fantastic. In these plains, in these valleys, in these
mountains, in these cities, labourers receive higher wages, merchants
and manufacturers make their fortunes faster, capitalists come into
contact with mightier interests, landlords draw higher rents from this
prosperity—all the sources of profit are more abundant than in Europe.
And these conditions have made it possible for a few of Fortune’s
favourites to pile up in the course of a single lifetime wealth whose
vastness makes the brain swim. America has, in fact, succeeded in
producing riches at a rate of speed that man has never yet attained
elsewhere in the world. She has been the principal factor in the
fabulous increase of the world’s wealth during the last fifty years. Her
riches have become one of the historic forces of our civilisation, and
one of the principal preoccupations of the European mind.
Whence come these vast riches and whither do they go? How is it
that America can grow rich so much faster than Europe? Is it thanks to
far more fortunate physical conditions, which bear no relation to the
deserts of man? Or is it in consequence of moral and intellectual
qualities which are lacking in Europeans? And what will be the ultimate
effect of this economic superiority? Riches may be the goal of an
individual’s efforts; for a nation they can only be means to conquer the
other good things of life which we call civilisation: glory, grandeur,
power, beauty, knowledge, moral refinement. Can America, and will
she, make use of her riches to rob Europe of the intellectual and moral
leadership which the latter still possesses? Or will these riches, too
swiftly won, exercise an evil influence simultaneously upon Europe and
America, by making both continents more materialistic?
* * * * *
Such is the riddle of America, which, for some time past, has been
steadily forcing itself upon the attention of Europe. To arrive at an
answer, we must know whether the influence of a too swift economic
development of the New World upon the higher activities of the mind,
upon morals, upon science, art, and religion is beneficial or the reverse.
The detractors of America—and there are many of them in Europe—
affirm without hesitation that the Americans are barbarians laden with
gold; that they think only of making money, and that, in consequence
of their riches, they lower the level of Europe’s ancient civilisation and
infect its beautiful traditions with a crass materialism. Admirers of
America, on the contrary—and of these there are as many in Europe as
there are detractors—will tell you that the New World is giving to the
Old a unique example of energy, activity, intelligence, and daring. Let
old Europe then give heed; beyond the Atlantic, young rivals are
girding themselves with new weapons to dispute with her the
superiority of which she is proud. What must one think of these
conflicting answers to the puzzle?
Let us begin with the reasoning of the detractors: “Americans are
barbarians laden with gold.” In order to simplify the discussion, let us
limit our examination to the United States, which is justly entitled to
represent contemporary America with all its qualities and all its defects.
No long sojourn within the borders of the United States is necessary to
convince a person that in the great Republic people think only of
making money. A writer partial to paradox might well amuse himself
with proving that the Americans are more idealistic than the
Europeans, or even that they are a mystical people. Anyone who cares
to find arguments to establish this thesis may well be embarrassed by
their number. For instance, would a people which despised the higher
activities of the mind have been able to create the philosophical
doctrine which is popularly known to us under the name of
“Pragmatism”? The Pragmatist affirms that all ideas capable of
rendering useful service are true. He takes utility as his standard of the
measure of truth. This theory has seemed to many writers of the Old
World a decisive proof of the practical mind of the American people,
who never forget their material interests, even in connection with
metaphysical questions. This, however, is a mistake. Pragmatism does
not propose to subordinate the ideal to practical interest. Its purpose is
to reconcile opposing doctrines by proving that all ideas, even those
which seem mutually exclusive, can help us to become wiser, stronger,
better. What service is there then in struggling to make one idea
triumph over another instead of allowing men to draw from each idea
the good which each can yield? In a word, Pragmatism, as America has
conceived it, is a mighty effort to give the right of expression in modern
civilisation to all religious and philosophical doctrines which in the past
have stained the world with their sanguinary struggles.
A beautiful doctrine this, which may lend itself to many objections;
but true or false, it proves that the people who have conceived it, far
from despising the ideal, have such respect for all ideas and all beliefs,
that they have not the courage to repel a single one. Such a people
wishes to learn all and understand all.
Another proof of this same characteristic is furnished by American
universities. Europeans have all heard descriptions of these great
American universities, Harvard and Columbia, for example. They are
true cities of learning with vast and splendid buildings, gardens,
pavilions, laboratories, museums, libraries, athletic fields for physical
exercises, pools where students can go to swim. They are enormously
rich and, at the same time, always in dire straits. How can that be?
Because no speciality or item of perfection is allowed to be lacking. All
the languages and the literatures of the world which have reached any
degree of importance, all the histories, all the sciences,—judicial, social,
moral, physical, natural,—all the divisions of mathematics, and all the
philosophies, are taught there by hundreds of professors. Private
citizens of the rich classes, bankers, manufacturers, merchants, have in
a great degree met from their private purses the steadily growing
needs of the universities.
There is the same tendency in art. That American cities are ugly, I
willingly admit. It would need much courage, no doubt, to brand this
affirmation as false, but it would also be unjust to deny that America is
making mighty efforts to beautify her cities. All the schools of
architecture in Europe, especially that of Paris, are full of Americans
hard at work. The sums which cities, states, banks, insurance
companies, universities, and railroads, have spent in beautifying their
magnificent edifices is fabulous. Not all these buildings, by any means,
are masterpieces, but there are many which are very beautiful. America
has architects of indisputable worth. In Europe, men like to repeat that
Americans buy at extravagant prices objects of ancient art, or things
that pass for such, not distinguishing those which are beautiful and
ancient from those which are inferior and counterfeit. But those who
have seen something of the houses of rich Americans know that,
although there are snobs and dupes in America, as everywhere else,
there are also people who know the meaning of art, who know how to
buy beautiful things, and who search the world over for them. You will
find in the streets of New York every variety of architecture, just as you
find in its libraries all the literatures of the world, and in its theatres all
the music, and in its houses all the decorative arts.
“The barbarian laden with gold” is, then, a legendary personage,
but it is not at all surprising that such a conception should exist.
Modern society is organised in such fashion that it is impossible even to
conceive of a people at once rich and ignorant. Industry, business,
agriculture, demand nowadays very special technical knowledge, and a
very complete social organisation; that is to say, they imply a scientific,
political, and judicial civilisation of a reasonably high order. Thus
America is not at all uninterested in the higher activities of the mind. It
would be more just to say that as a nation, and without regard to
individual instances, she interests herself in such activities less than in
industry, in business, and in agriculture. But is not this also the case
with Europe? Who would dare affirm that the progress of the arts and
sciences and letters is at this moment the principal concern of the
governments and of the influential classes of the Old World? We
Europeans have only to listen to what people round about us are
saying. Their talk is all of bringing the cultivation of the land to
economic perfection, of opening coal and iron mines, of harnessing
waterfalls, of developing industries, of increasing exports. Kings who
rule “by the grace of God” publicly declare that nothing interests them
so much as the business of their countries!
If all this were characteristic only of American barbarism, we should
be obliged to admit that Europe is Americanising herself with
disconcerting rapidity. But this economic effort of Europe in turn
presents nothing that need surprise us; like the American development,
it is only the dizzy acceleration of a vast historic movement whose
beginnings go back to the far distant day when an obscure and
obstinate Genoese set sail, and in the midst of the waters of the
Atlantic crossed the impassable boundary of the Old World. Yes, before
that day, Europe had created admirable arts and literatures, profound
philosophies, consoling religions, lofty morals, wise systems of justice,
but—she was poor. She produced little, and produced it slowly; she had
defied tradition and authority; she had fettered human energy by a
multitude of laws, precepts, and prejudices. To humble men’s pride,
she kept repeating to them that they were feeble and corrupt
creatures. She taught them to use Virgil’s beautiful figure that they
were like “a rower who painfully forces his boat against the current of
the stream. Evil be on his head if for one instant he forgets, and ceases
to struggle against the current’s force; in that moment, he is lost; the
flood sweeps away his fragile boat.”
One fine day, however, Europe discovered a vast continent in the
midst of the ocean. Then it dawned upon her that Prometheus had
been but a clumsy thief, for he had stolen only a tiny spark of fire; she
discovered mines, coal, and electricity. She created the steam-engine
and all the other machines which have been derived from it. She
succeeded in multiplying riches with a rapidity unimagined by remoter
ancestors. From that moment, man no longer contented himself with
dreaming of the Promised Land. He wished to go there. He destroyed
all the traditions, the laws, and institutions which place limitations upon
the store of human energy. He learned to work swiftly. At a single
stroke, he conquered liberty and riches, and he conceived the idea of
progress. If America seems to-day to symbolise this movement, which
has turned the world topsy-turvy, the movement was derived from
Europe. After having conceived the idea of such a revolution, could
Europe remain untouched by it?
* * * * *
It would appear then that the riddle of America is very simple. The
answer contains nothing to make us uneasy. The riches of the New
World threaten no catastrophe to the noblest traditions of our
civilisation. For New York’s wealth is only a part of the riches produced
in the same economic development in the two worlds. The ultimate
development of these mighty riches might be merely a general
advance, both material and ideal, of Europe and America. Rich and
prosperous Americans might try to assimilate the culture of Europe,
and on her part Europe, in her effort to increase her own riches, might
seek to equal America. But a historian of antiquity who returns from
America cannot share this optimism. In the lap of modern civilisation,
there are twin worlds struggling with each other for leadership. But
these two worlds are not, as people are apt to think, Europe and
America. Their names are Quality and Quantity.
The civilisations from which our own is sprung were poor indeed.
They set limits to their desires, their ambitions, their spirit of initiative,
their audacity, their originality. They brought forth slowly and a little at
a time, and suffered continuously from the insufficiency of their
material resources. They looked upon the amassing of wealth merely
as a painful necessity; but, in all things, they sought to attain the
difficult model of perfection, whether in art, or in literature, or in the
realms of morality and religion. The aristocratic character of almost all
the industries of the past, the importance which was formerly bestowed
on the decorative arts and on all questions of personal morality,
ceremonial, and form—these are all proofs of it. It was Quality, not
Quantity, which carried our forefathers forward. All the limitations to
which these civilisations were subject, so astonishing to us to-day, were
only the necessary cost of these perfections which men once so
ardently desired. We have turned upside down the world our ancestors
lived in. We have made our goal the multiplication of riches. We have
won liberty, but we have been obliged to abandon almost all the
ancient ideals of perfection, sacrificing Quality in everything.
How many of the difficulties which torture this brilliant period of
ours so cruelly are the result of this duel between Quality and Quantity!
Look, for example, at the present crisis in the study of the classics.
Why did men formerly study Homer and Cicero with passionate zeal?
Because, in those days, the great Greek and Latin writers were the
models of that literary perfection, so greatly admired by the influential
classes, which was not merely an ornament of the mind. The
attainment of perfection often carried with it the admiration of the
public, fame, sometimes even glory and high rank. In this last century,
however, these models have lost much of their prestige, either on
account of the multitude of literatures which have come to be known
and liked, or because they have proved troublesome to a period
compelled to write too much and too quickly. Just imagine a candidate
for the presidency of the United States who should pronounce ten or
fifteen long orations daily and who should in each discourse show
himself the perfect orator according to the rules of Cicero or Quintilian!
The day when classical culture ceased to be an official school of literary
taste, on that day it was condemned to die; and scientific philology,
which we have sought to set up in its place, can only serve to bury its
corpse. No longer models for posterity, the books of the ancient
authors have become like any others, and are less interesting for the
majority of readers than the works of modern literatures.
It is the fashion nowadays to discuss the crisis which threatens all
the arts. We must, however, remember to preserve a distinction. We
must divide the arts into two categories: those which serve to amuse
men by helping them to pass the time agreeably, like music, the

Download Full Multimedia Security Watermarking Steganography and Forensics 1st Edition Frank Y. Shih (Editor) PDF All Chapters

Uploaded by

Copyright:

Available Formats

Download Full Multimedia Security Watermarking Steganography and Forensics 1st Edition Frank Y. Shih (Editor) PDF All Chapters

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Download Full Multimedia Security Watermarking Steganography and Forensics 1st Edition Frank Y. Shih (Editor) PDF All Chapters

Uploaded by

Copyright:

Available Formats

Download the full version of the ebook at

Multimedia Security Watermarking

Explore and download more ebook at https://ebookultra.com

Disappearing cryptography Information hiding Steganography

Multimedia Security Handbook 1st Edition Borko Furht

Android Forensics Investigation Analysis and Mobile

Cyber Security Understanding Cyber Crimes Computer

The Facebook era 4. print Edition Shih

Exploratory Image Databases AP Communications Networking

Learn Chinese from modern writers C. W. Shih

Microbial Forensics 1st Edition Roger G. Breeze

Giving readers an in-depth overview of different aspects of information

Presented case studies cover:

• The execution (and feasibility) of techniques used to discover hidden

6000 Broken Sound Parkway, NW

CRC Press is an imprint of the

© 2013 by Taylor & Francis Group, LLC

No claim to original U.S. Government works

International Standard Book Number-13: 978-1-4398-7332-8 (eBook - PDF)

and the CRC Press Web site at

Part I Multimedia Mining and Classification

Chapter 1 Multimedia Duplicate Mining toward Knowledge Discovery..............3

Chapter 2 Discriminative Learning-Assisted Video Semantic Concept

Chapter 3 Improved Feature Vocabulary-Based Method for Image

Chapter 5 Digital Watermarking Based on Chaotic Map and Reference

Chapter 6 Pseudo-Random Pixel Rearrangement Algorithm Based on

Chapter 7 Reversible Data-Hiding Techniques for Digital Images................... 115

Chapter 8 Watermarking Based on Local Binary Pattern Operators................ 141

Chapter 9 Authentication of JPEG Images Based on Genetic Algorithms....... 165

Chapter 10 An Efficient Block-Based Fragile Watermarking System

Part III Steganography

Chapter 11 Survey of Image Steganography and Steganalysis............................ 201

Chapter 12 Digital Steganographic Schemes Based on Image Vector

Chapter 13 Differential Evolution-Based Algorithm for Breaking

Chapter 14 Genetic Algorithm-Based Methodology for Breaking the

Chapter 15 Image Inpainting Using an Enhanced Exemplar-Based

Chapter 16 A Comparison Study on Copy–Cover Image Forgery Detection..... 297

Chapter 17 Chaos-Based Hash Function with Both Modification

Chapter 18 Video Forensics................................................................................. 327

Chapter 19 Using the Self-Synchronizing Method to Improve Security

Chapter 20 Behavior Modeling of Human Objects in Multimedia Content........ 357

Overview of the Book

random sampling procedure together with an increased number of sample points is

methodology by adjusting gray values of a cover-image while creating the desired

Features of the Book

Feedback on the Book

center in the university. In 2004, he was a visiting scholar at Princeton University.

Alexander Sheppard received a BS in computer science from the New Jersey

professor in the Department of Computer Science and Information Engineering,

Hong-Ren Su received his BS in biomedical imaging and radiological science from

Hsiao-Rong Tyan received her BS in electronic engineering from Chung Yuan

Boris S. Verkhovsky is a professor in the Computer Science Department at the New

Zhi-Hui Wang received a BS in software engineering in 2004 from the North

Xiaomeng Wu is a project researcher at the National Institute of Informatics, Japan.

retrieval, and near-duplicate detection. He is a member of the IEEE Computer

Yi-Ta Wu received his BS in physics from Tamkang University, Taipei, Taiwan, in

Di Xiao received a BS from Sichuan University, Chengdu, China in 1997. He

Wenyin Zhang is an assistant professor in the School of Informatics, Linyi

s­ calability issue of a near-duplicate search. Finally, a discussion on these techniques

1.2 Selection Criterion of Duplicate Mining Methods

s calability issue of a near-duplicate search. Finally, a discussion on these techniques

types of chronological distributions involving five types of information including