Download the full version of the textbook now at textbookfull.

com

Data Analytics and Learning Proceedings of

DAL 2018 P. Nagabhushan

https://textbookfull.com/product/data-analytics-
and-learning-proceedings-of-
dal-2018-p-nagabhushan/

Explore and download more textbook at https://textbookfull.com

 Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.

Data Analytics and Learning: Proceedings of DAL 2022

(Lecture Notes in Networks and Systems, 779) 1st Edition
P. Nagabhushan
https://textbookfull.com/product/data-analytics-and-learning-
proceedings-of-dal-2022-lecture-notes-in-networks-and-systems-779-1st-
edition-p-nagabhushan/
textbookfull.com

Recent Developments in Machine Learning and Data Analytics

IC3 2018 Jugal Kalita

https://textbookfull.com/product/recent-developments-in-machine-
learning-and-data-analytics-ic3-2018-jugal-kalita/

textbookfull.com

Data Management Analytics and Innovation Proceedings of

ICDMAI 2018 Volume 2 Valentina Emilia Balas

https://textbookfull.com/product/data-management-analytics-and-
innovation-proceedings-of-icdmai-2018-volume-2-valentina-emilia-balas/

textbookfull.com

Software Architecture for Big Data and the Cloud 1st

Edition Ivan Mistrik

https://textbookfull.com/product/software-architecture-for-big-data-
and-the-cloud-1st-edition-ivan-mistrik/

textbookfull.com
Transforming Organizations Through Flexible Systems
Management P.K. Suri

https://textbookfull.com/product/transforming-organizations-through-
flexible-systems-management-p-k-suri/

textbookfull.com

Small business management launching and growing new

ventures Sixth Canadian Edition Longenecker

https://textbookfull.com/product/small-business-management-launching-
and-growing-new-ventures-sixth-canadian-edition-longenecker/

textbookfull.com

Pro Microsoft HDInsight Hadoop on Windows 1st Edition

Debarchan Sarkar (Auth.)

https://textbookfull.com/product/pro-microsoft-hdinsight-hadoop-on-
windows-1st-edition-debarchan-sarkar-auth/

textbookfull.com

Intermittent Fasting for Women Over 50 1st Edition

Jennifer Robinson

https://textbookfull.com/product/intermittent-fasting-for-women-
over-50-1st-edition-jennifer-robinson/

textbookfull.com

ERS Practical Handbook of Invasive Mechanical Ventilation

Leo Heunks

https://textbookfull.com/product/ers-practical-handbook-of-invasive-
mechanical-ventilation-leo-heunks/

textbookfull.com
The Book of Delights: Essays 1st Edition Ross Gay

https://textbookfull.com/product/the-book-of-delights-essays-1st-
edition-ross-gay/

textbookfull.com
Lecture Notes in Networks and Systems 43

P. Nagabhushan
D. S. Guru
B. H. Shekar
Y. H. Sharath Kumar Editors

Data
Analytics and
Learning
Proceedings of DAL 2018
Lecture Notes in Networks and Systems

Volume 43

Series editor
Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland
e-mail: [email protected]
The series “Lecture Notes in Networks and Systems” publishes the latest
developments in Networks and Systems—quickly, informally and with high quality.
Original research reported in proceedings and post-proceedings represents the core
of LNNS.
Volumes published in LNNS embrace all aspects and subfields of, as well as
new challenges in, Networks and Systems.
The series contains proceedings and edited volumes in systems and networks,
spanning the areas of Cyber-Physical Systems, Autonomous Systems, Sensor
Networks, Control Systems, Energy Systems, Automotive Systems, Biological
Systems, Vehicular Networking and Connected Vehicles, Aerospace Systems,
Automation, Manufacturing, Smart Grids, Nonlinear Systems, Power Systems,
Robotics, Social Systems, Economic Systems and other. Of particular value to both
the contributors and the readership are the short publication timeframe and the
world-wide distribution and exposure which enable both a wide and rapid
dissemination of research output.
The series covers the theory, applications, and perspectives on the state of the art
and future developments relevant to systems and networks, decision making, control,
complex processes and related areas, as embedded in the fields of interdisciplinary
and applied sciences, engineering, computer science, physics, economics, social, and
life sciences, as well as the paradigms and methodologies behind them.

Advisory Board
Fernando Gomide, Department of Computer Engineering and Automation—DCA, School of
Electrical and Computer Engineering—FEEC, University of Campinas—UNICAMP,
São Paulo, Brazil
e-mail: [email protected]
Okyay Kaynak, Department of Electrical and Electronic Engineering, Bogazici University,
Istanbul, Turkey
e-mail: [email protected]
Derong Liu, Department of Electrical and Computer Engineering, University of Illinois
at Chicago, Chicago, USA and Institute of Automation, Chinese Academy of Sciences,
Beijing, China
e-mail: [email protected]
Witold Pedrycz, Department of Electrical and Computer Engineering, University of Alberta,
Alberta, Canada and Systems Research Institute, Polish Academy of Sciences, Warsaw,
Poland
e-mail: [email protected]
Marios M. Polycarpou, KIOS Research Center for Intelligent Systems and Networks,
Department of Electrical and Computer Engineering, University of Cyprus, Nicosia, Cyprus
e-mail: [email protected]
Imre J. Rudas, Óbuda University, Budapest Hungary
e-mail: [email protected]
Jun Wang, Department of Computer Science, City University of Hong Kong
Kowloon, Hong Kong
e-mail: [email protected]

More information about this series at http://www.springer.com/series/15179

P. Nagabhushan D. S. Guru
•

B. H. Shekar Y. H. Sharath Kumar

•

Editors

Data Analytics and Learning

Proceedings of DAL 2018

123
Editors
P. Nagabhushan B. H. Shekar
Indian Institute of Information Technology Department of Studies in Computer Science
Allahabad, Uttar Pradesh, India Mangalore University
Mangalore, Karnataka, India
D. S. Guru
Department of Computer Science Y. H. Sharath Kumar
and Engineering, CBCS Education Department of Information Science
University of Mysore and Engineering
Mysuru, Karnataka, India Maharaja Institute of Technology
Belawadi, Karnataka, India

ISSN 2367-3370 ISSN 2367-3389 (electronic)

Lecture Notes in Networks and Systems
ISBN 978-981-13-2513-7 ISBN 978-981-13-2514-4 (eBook)
https://doi.org/10.1007/978-981-13-2514-4

Library of Congress Control Number: 2018954017

© Springer Nature Singapore Pte Ltd. 2019

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface

We write this message with deep satisfaction to the proceedings of the “First
International Conference on Data Analytics and Machine Learning 2018
(DAL 2018)” held on March 30 and 31, 2018, at Mysore, Karnataka, India, which
has the central theme “Data Analytics and its Application.” Our research experi-
ences in related areas for the last decade have inspired us to conduct DAL 2018.
This conference was planned to provide a platform for researchers from both
academia and industries where they can discuss and exchange their research
thoughts to have better future research plans, particularly in the fields of data
analytics and machine learning. Soon after we notified a call for original research
papers, there has been a tremendous response from the researchers. There were 150
papers submitted, out of which we could accommodate only 50 papers based on the
reports of the reviewers. Each paper was blindly reviewed by at least two experts
from the related areas. The overall acceptance rate is about 30 %. The conference is
aimed at image processing, signal processing, pattern recognition, document pro-
cessing, biomedical processing, computer vision, biometrics, data mining and
knowledge discovery, information retrieval and information coding. For all these
areas, we got a number of papers reflecting their right combinations. I hope that the
readers will appreciate and enjoy the papers published in the proceedings. We could
make this conference a successful one, though it was launched at a relatively short
notice. It was because of the good response from the research community and the
good effort put in by the reviewers to support us with timely reviews. The authors of
all the papers submitted deserve our acknowledgments. The proceedings are pub-
lished and indexed by Springer-LNEE, which is known for bringing out this type of
proceedings. Special thanks to them.

v
vi Preface

We would also like to thank the help of EasyChair in the submission, review,
and proceedings creation processes. We are very pleased to express our sincere
thanks to Springer, especially Jayanthi Narayanaswamy, Jayarani Premkumar,
Aninda Bose, and the editorial staff, for their support in publishing the proceedings
of DAL 2018.

Allahabad, India Dr. P. Nagabhushan

Mysuru, India Dr. D. S. Guru
Mangalore, India Dr. B. H. Shekar
Belawadi, India Dr. Y. H. Sharath Kumar
Contents

Recognition of Seven-Segment Displays from Images

of Digital Energy Meters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Thotreingam Kasar
An Enhanced Task Scheduling in Cloud Computing Based
on Hybrid Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Mokhtar A. Alworafi, Atyaf Dhari, Sheren A. El-Booz, Aida A. Nasr,
Adela Arpitha and Suresha Mallappa
A Moment-Based Representation for Online Telugu Handwritten
Character Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Krishna Chaithanya Movva and Viswanath Pulabaigari
A Switch-Prioritized Load-Balancing Technique in SDN . . . . . . . . . . . . 39
K. A. Vani, J. Prathima Mabel and K. N. Rama Mohan Babu
A Robust Human Gait Recognition Approach
Using Multi-interval Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
V. G. Manjunatha Guru and V. N. Kamalesh
Keyword Spotting in Historical Devanagari Manuscripts
by Word Matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
B. Sharada, S. N. Sushma and Bharathlal
A Parallel Programming Approach for Estimation of Depth
in World Coordinate System Using Single Camera . . . . . . . . . . . . . . . . 77
C. Rashmi and G. Hemantha Kumar
An Alternate Voltage-Controlled Current Source for Electrical
Impedance Tomography Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Venkatratnam Chitturi and Nagi Farrukh

vii
viii Contents

Analyzing and Comparison of Movie Rating Using Hadoop

and Spark Frame Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Akshaya Devadiga, C. V. Aravinda and H. N. Prakash
Classification of Osteoarthritis-Affected Images Based on Edge
Curvature Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Ravindra S. Hegadi, Trupti D. Pawar and Dattatray I. Navale
Identification of Knee Osteoarthritis Using Texture Analysis . . . . . . . . . 121
Ravindra S. Hegadi, Umesh P. Chavan and Dattatray I. Navale
Logo Retrieval and Document Classification Based
on LBP Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
C. Veershetty and Mallikarjun Hangarge
Semantic Relatedness Measurement from Wikipedia and WordNet
Using Modified Normalized Google Distance . . . . . . . . . . . . . . . . . . . . . 143
Saket Karve, Vasisht Shende and Swaroop Hople
Design and Simulation of Neuro-Fuzzy Controller for Indirect
Vector-Controlled Induction Motor Drive . . . . . . . . . . . . . . . . . . . . . . . . 155
B. T. Venu Gopal and E. G. Shivakumar
Entropy-Based Approach for Enabling Text Line Segmentation
in Handwritten Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
G. S. Sindhushree, R. Amarnath and P. Nagabhushan
Automated Parameter-Less Optical Mark Recognition . . . . . . . . . . . . . . 185
N. C. Dayananda Kumar, K. V. Suresh and R. Dinesh
A Bimodal Biometric System Using Palmprint and Face Modality . . . . 197
N. Harivinod and B. H. Shekar
Circular Map Pattern Spectrum—An Accurate Descriptor
for Shape Representation and Classification . . . . . . . . . . . . . . . . . . . . . . 207
Bharathi Pilar and B. H. Shekar
Features Fusion for Retrieval of Flower Videos . . . . . . . . . . . . . . . . . . . 221
D. S. Guru, V. K. Jyothi and Y. H. Sharath Kumar
Offline Signature Verification: An Approach Based
on User-Dependent Features and Classifiers . . . . . . . . . . . . . . . . . . . . . . 235
K. S. Manjunatha, H. Annapurna and D. S. Guru
Simple DFA Construction Algorithm Using Divide-and-Conquer
Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Darshan D. Ruikar and Ravindra S. Hegadi
Estimating the Rating of the Reviews Based on the Text . . . . . . . . . . . . 257
Mohammadamir Kavousi and Sepehr Saadatmand
Contents ix

Multimodal Biometric Recognition System Based

on Nonparametric Classifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
H. D. Supreetha Gowda, G. Hemantha Kumar and Mohammad Imran
Classification of Multi-class Microarray Cancer Data Using Ensemble
Learning Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
B. H. Shekar and Guesh Dagnew
A Study of Applying Different Term Weighting Schemes on Arabic
Text Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
D. S. Guru, Mostafa Ali, Mahamad Suhil and Maryam Hazman
A Survey on Different Visual Speech Recognition Techniques . . . . . . . . 307
Shabina Bhaskar, T. M. Thasleema and R. Rajesh
Activity Recognition from Accelerometer Data Using Symbolic
Data Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
P. G. Lavanya and Suresha Mallappa
Automated IT Service Desk Systems Using Machine Learning
Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
S. P. Paramesh and K. S. Shreedhara
Automatic Segmentation and Breast Density Estimation for Cancer
Detection Using an Efficient Watershed Algorithm . . . . . . . . . . . . . . . . . 347
Tejas Nayak, Nikitha Bhat, Vikram Bhat, Sannidhi Shetty,
Mohammed Javed and P. Nagabhushan
Offline Signature Verification Based on Partial Sum of Second-Order
Taylor Series Expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
B. H. Shekar, Bharathi Pilar and D. S. Sunil Kumar
Urban LULC Change Detection and Mapping Spatial Variations
of Aurangabad City Using IRS LISS-III Temporal Datasets and
Supervised Classification Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
Ajay D. Nagne, Amol D. Vibhute, Rajesh K. Dhumal, Karbhari V. Kale
and S. C. Mehrotra
Study of Meta-Data Enrichment Methods to Achieve
Near Real Time ETL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
N. Mohammed Muddasir and K. Raghuveer
BornBaby Model for Software Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . 403
H. L. Gururaj and B. Ramesh
Lung Cancer Detection Using CT Scan Images: A Review
on Various Image Processing Techniques . . . . . . . . . . . . . . . . . . . . . . . . 413
A. P. Ayshath Thabsheera, T. M. Thasleema and R. Rajesh
Visit https://textbookfull.com
now to explore a rich
collection of eBooks, textbook
and enjoy exciting offers!
x Contents

Deep Learning Approach for Classification of Animal Videos . . . . . . . . 421

N. Manohar, Y. H. Sharath Kumar, G. Hemantha Kumar and Radhika Rani
Building Knowledge Graph Based on User Tweets . . . . . . . . . . . . . . . . . 433
Seyedmahmoud Talebi, K. Manoj and G. Hemantha Kumar
Interoperability and Security Issues of IoT in Healthcare . . . . . . . . . . . 445
M. Shankar Lingam and A. M. Sudhakara
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
Editors and Contributors

About the Editors

P. Nagabhushan is Director of the Indian Institute of Information Technology,

Allahabad, Uttar Pradesh. He has previously served as Professor at the University
of Mysore and Amrita University. His areas of specialization are pattern recogni-
tion, digital image analysis and processing, document image analysis, and knowl-
edge mining. He has published more than 500 papers in peer-reviewed journals and
conferences in these areas. He has received research funding from various orga-
nizations, e.g., DRDO, MHRD, AICTE, ICMR, UGC, UGC, UPE, and SAP, and
has more than 30 years of teaching experience.

D. S. Guru is Professor in the Department of Studies in Computer Science,

University of Mysore. He completed his doctorate in computer science from the
University of Mysore in 2000 and has published more than 200 research papers in
national and international journals and conference proceedings. He has served as
Program Chair for the 2009 International Conference on Signal and Image
Processing in Mysore and is currently Reviewer for the International Journal of
Pattern Recognition Letters, Journal of Pattern Recognition, Journal of Document
Analysis and Recognition, Journal of Image and Vision Computing, and IEEE
Transactions on Neural Networks.

B. H. Shekar is Professor in the Department of Studies in Computer Science,

Mangalore University, India. His research interests include pattern recognition,
image processing, database systems, and algorithms. He is an active researcher and
has published more than 100 papers in peer-reviewed journals and conferences. He
was granted a DST-DAAD Fellowship in 2002 and was the principal investigator
for many research projects funded by DST-RFBR and UGC. He is also a member
of many academic societies, e.g., the International Association for Pattern
Recognition, IACSIT, and Indian Unit for Pattern Recognition and Artificial
Intelligence.

xi
xii Editors and Contributors

Y. H. Sharath Kumar is Associate Professor in the Department of Information

Science and Engineering, Maharaja Institute of Technology, Mysuru. He received
his M.Tech. in 2009 and his Ph.D. in 2014, both from the University of Mysore. His
research interests include image and video processing, pattern recognition, Sign
Language Video Analysis, logo classification, biometrics, and precision agriculture.
He has authored a number of papers in reputed journals and conference proceedings.

Contributors

Mostafa Ali Department of Studies in Computer Science, University of Mysore,

Manasagangotri, Mysuru, Karnataka, India
Mokhtar A. Alworafi DoS in Computer Science, University of Mysore, Mysuru,
India
R. Amarnath Department of Studies in Computer Science, University of Mysore,
Mysuru, Karnataka, India
H. Annapurna Department of Studies in Computer Science, University of
Mysore, Mysuru, Karnataka, India
C. V. Aravinda NMAM Institute of Technology NITTE, Karakal, Karnataka,
India
Adela Arpitha DoS in Computer Science, University of Mysore, Mysuru, India
A. P. Ayshath Thabsheera Department of Computer Science, Central University
of Kerala, Kasaragod, Kerala, India
Bharathlal Department of Studies in Computer Science, University of Mysore,
Mysuru, India
Shabina Bhaskar Central University of Kerala, Kasaragod, Kerala, India
Nikitha Bhat Department of CSE, NMAM Institute of Technology, Nitte, India
Vikram Bhat Department of CSE, NMAM Institute of Technology, Nitte, India
Umesh P. Chavan School of Computational Sciences, Solapur University,
Solapur, India
Venkatratnam Chitturi Department of Instrumentation Technology,
GSSSIETW, Mysuru, India
Guesh Dagnew Department of Computer Science, Mangalore University,
Karnataka, Mangalore, India
N. C. Dayananda Kumar Department of Electronics and Communication
Engineering, Siddaganga Institute of Technology, Tumkur, India
Editors and Contributors xiii

Akshaya Devadiga NMAM Institute of Technology NITTE, Karakal, Karnataka,

India
Atyaf Dhari College Education for Pure Science, Thi Qar University, Nasiriyah,
Iraq
Rajesh K. Dhumal Department of CS and IT, Dr. Babasaheb Ambedkar
Marathwada University, Aurangabad, Maharashtra, India
R. Dinesh Department of Information Science and Engineering, Jain University,
Bengaluru, India
Sheren A. El-Booz Department of Computer Science and Engineering, Menoufia
University, Shibin El Kom, Egypt
Nagi Farrukh Faculty of Mechanical Engineering, Universiti Tenaga Nasional,
Kajang, Selangor, Malaysia
D. S. Guru Department of Studies in Computer Science, University of Mysore,
Manasagangotri, Mysuru, India
H. L. Gururaj Vidyavardhaka College of Engineering, Mysuru, India
Mallikarjun Hangarge Department of Computer Science, Karnatak Arts Science
and Commerce College, Bidar, India
N. Harivinod Department of Computer Science and Engineering, Vivekananda
College of Engineering and Technology, Puttur, Karnataka, India
Maryam Hazman Central Laboratory for Agricultural Experts Systems,
Agricultural Research Center (ARC), Giza, Egypt
Ravindra S. Hegadi School of Computational Sciences, Solapur University,
Solapur, India
G. Hemantha Kumar Department of Computer Science, University of Mysore,
Mysuru, India; High Performance Computing Project, DoS in CS, University of
Mysore, Mysuru, India
Swaroop Hople Veermata Jijabai Technological Institute, Mumbai, India
Mohammad Imran Department of Computer Science, University of Mysore,
Mysuru, India
Mohammed Javed Department of IT, Indian Institute of Information Technology,
Allahabad, Allahabad, India
V. K. Jyothi Department of Studies in Computer Science, University of Mysore,
Manasagangotri, Mysuru, India
Karbhari V. Kale Department of CS and IT, Dr. Babasaheb Ambedkar
Marathwada University, Aurangabad, Maharashtra, India
xiv Editors and Contributors

V. N. Kamalesh VTU RRC, Belagavi, Karnataka, India; TJIT, Bengaluru, India

Saket Karve Veermata Jijabai Technological Institute, Mumbai, India
Thotreingam Kasar Kaaya Tech Inc., Mysuru, India
Mohammadamir Kavousi Southern Illinois University, Carbondale, USA
P. G. Lavanya DoS in Computer Science, University of Mysore, Mysuru, India
Suresha Mallappa DoS in Computer Science, University of Mysore, Mysuru,
India
V. G. Manjunatha Guru GFGC, Honnāli, India; VTU RRC, Belagavi,
Karnataka, India
K. S. Manjunatha Maharani’s Science College for Women, Mysuru, Karnataka,
India
N. Manohar Department of Computer Science, University of Mysore, Mysuru,
India
K. Manoj Bengaluru, India
S. C. Mehrotra Department of CS and IT, Dr. Babasaheb Ambedkar Marathwada
University, Aurangabad, Maharashtra, India
N. Mohammed Muddasir Department of Information Science and Engineering,
VVCE, Mysuru, India; NIE, Mysuru, India
Krishna Chaithanya Movva Indian Institute of Information Technology,
Sri City, Chittoor, Andhra Pradesh, India
P. Nagabhushan Department of IT, Indian Institute of Information Technology,
Allahabad, Allahabad, India
Ajay D. Nagne Department of CS and IT, Dr. Babasaheb Ambedkar Marathwada
University, Aurangabad, Maharashtra, India
Aida A. Nasr Department of Computer Science and Engineering, Menoufia
University, Shibin El Kom, Egypt
Dattatray I. Navale School of Computational Sciences, Solapur University,
Solapur, India
Tejas Nayak Department of CSE, NMAM Institute of Technology, Nitte, India
S. P. Paramesh Department of Studies in Computer Science and Engineering,
U.B.D.T College of Engineering, Davanagere, Karnataka, India
Trupti D. Pawar School of Computational Sciences, Solapur University, Solapur,
India
Bharathi Pilar Department of Computer Science, University College, Mangalore,
Karnataka, India
Editors and Contributors xv

H. N. Prakash Rajeev Institute of Technology, Hassan, Karnataka, India

J. Prathima Mabel Department of Information Science and Engineering,
Dayananda Sagar College of Engineering, Bengaluru, India
Viswanath Pulabaigari Indian Institute of Information Technology, Sri City,
Chittoor, Andhra Pradesh, India
K. Raghuveer Department of Information Science and Engineering, NIE, Mysuru,
India
R. Rajesh Department of Computer Science, Central University of Kerala,
Kasaragod, Kerala, India
K. N. Rama Mohan Babu Department of Information Science and Engineering,
Dayananda Sagar College of Engineering, Bengaluru, India
B. Ramesh Malnad College of Engineering, Hassan, India
Radhika Rani SBRR Mahajana First Grade College, Mysuru, India
C. Rashmi High Performance Computing Project, DoS in CS, University of
Mysore, Mysuru, India
Darshan D. Ruikar School of Computational Sciences, Solapur University,
Solapur, India
Sepehr Saadatmand Southern Illinois University, Carbondale, USA
M. Shankar Lingam University of Mysore, Manasagangotri, Mysuru, India
B. Sharada Department of Studies in Computer Science, University of Mysore,
Mysuru, India
Y. H. Sharath Kumar Department of Computer Science & Engineering and
Department of Information Science, Maharaja Institute of Technology, Mandya,
Karnataka, India
B. H. Shekar Department of Computer Science, Mangalore University,
Mangalore, Karnataka, India
Vasisht Shende Veermata Jijabai Technological Institute, Mumbai, India
Sannidhi Shetty Department of CSE, NMAM Institute of Technology, Nitte,
India
E. G. Shivakumar Department of Electrical Engineering, UVCE, Bangalore
University, Bengaluru, India
K. S. Shreedhara Department of Studies in Computer Science and Engineering,
U.B.D.T College of Engineering, Davanagere, Karnataka, India
G. S. Sindhushree Department of Studies in Computer Science, University of
Mysore, Mysuru, Karnataka, India
xvi Editors and Contributors

A. M. Sudhakara University of Mysore, Manasagangotri, Mysuru, India

Mahamad Suhil Department of Studies in Computer Science, University of
Mysore, Manasagangotri, Mysuru, Karnataka, India
D. S. Sunil Kumar Department of Computer Science, Mangalore University,
Mangalore, Karnataka, India
H. D. Supreetha Gowda Department of Computer Science, University of Mysore,
Mysuru, India
K. V. Suresh Department of Electronics and Communication Engineering,
Siddaganga Institute of Technology, Tumkur, India
S. N. Sushma Department of Studies in Computer Science, University of Mysore,
Mysuru, India
Seyedmahmoud Talebi Department of Computer Science, University of Mysore,
Mysuru, India
T. M. Thasleema Department of Computer Science, Central University of Kerala,
Kasaragod, Kerala, India
K. A. Vani Department of Information Science and Engineering, Dayananda
Sagar College of Engineering, Bengaluru, India
C. Veershetty Department of Computer Science, Gulbarga University,
Kalaburagi, India
B. T. Venu Gopal Department of Electrical Engineering, UVCE, Bangalore
University, Bengaluru, India
Amol D. Vibhute Department of CS and IT, Dr. Babasaheb Ambedkar
Marathwada University, Aurangabad, Maharashtra, India
Recognition of Seven-Segment Displays
from Images of Digital Energy Meters

Thotreingam Kasar

Abstract This paper describes a method to localize and recognize seven-segment

displays on digital energy meters. Color edge detection is first performed on a camera-
captured image of the device which is then followed by a run-length technique to
detect horizontal and vertical lines. The region of interest circumscribing the LCD
panel is determined based on the attributes of intersecting horizontal and vertical
lines. The extracted display region is preprocessed using the morphological black-
hat operation to enhance the text strokes. Adaptive thresholding is then performed
and the digits are segmented based on stroke features. Finally, the segmented digits
are recognized using a support vector machine classifier trained on a set of syntactic
rules defined for the seven-segment font. The proposed method can handle images
exhibiting uneven illumination, the presence of shadows, poor contrast, and blur, and
yields a recognition accuracy of 97% on a dataset of 175 images of digital energy
meters captured using a mobile camera.

Keywords Seven-segment displays · Character recognition

Camera-based document image analysis

1 Introduction

The camera provides a great opportunity for input from the physical world. In recent
years, it has become hard to define the term document due to the blurring in the
distinction between documents and user interfaces. In addition to imaging hard
copy documents, cameras are now increasingly being used to capture text present
on 3-D real-world objects such as buildings, billboards, road signs, license plates,
black/whiteboards, household appliances, or even on a T-shirt which otherwise would
be inaccessible to conventional scanner-based optical character recognition (OCR)
systems. Pervasive use of handheld digital cameras has immense potential for newer
applications that go far beyond what traditional OCR has to offer [1]. Recognizing

T. Kasar (B)
Kaaya Tech Inc., Mysuru 570017, India
e-mail: [email protected]

© Springer Nature Singapore Pte Ltd. 2019 1

P. Nagabhushan et al. (eds.), Data Analytics and Learning,
Lecture Notes in Networks and Systems 43,
https://doi.org/10.1007/978-981-13-2514-4_1
2 T. Kasar

Fig. 1 Typical images of seven-segment displays on digital energy meters exhibiting uneven light-
ing, shadows, poor contrast, blur, and presence of highlights occluding parts of a digit

text in real-world scenes can be considered as an extension of current OCR technol-

ogy widely available in the market. The unconstraint mode of document acquisition as
well the difference in the target document type calls for a new level of processing for
camera-captured images. While it may seem that the low variability of seven-segment
displays should make them easy to read, its automatic detection and recognition is
in fact a challenging task. A typical image of the display panel on an electronic
device contains mostly background clutter and other irrelevant texts. Therefore, a
preprocessing step of locating the display area is required before attempting to rec-
ognize text in the acquired image. Once the text region is localized, the subsequent
recognition task can be performed only on the detected region of interest so as to
obviate the effect of background clutter. However, the available technology is still
far from being able to reliably separate text from the background clutter. In addition,
images of LCD displays often exhibit poor contrast, blur, and may contain highlights
and specular reflections from the display surface which make them hard to segment.
Figure 1 shows some of these challenges commonly encountered in images of LCD
displays.
Recognition of Seven-Segment Displays from Images of Digital Energy Meters 3

2 Review of Related Work

While there are a lot of works on recognizing text from natural images [2–4], there
has been relatively less work that address the specific problem of recognizing seven-
segment displays on electronic devices. The Clearspeech system [5] requires special
markers to be affixed to the device to guide the system in localizing the display panel.
Shen and Coughlan [6] introduced a phone-based LED/LCD display reader, which
do not have such modification of the display. They employ horizontal and vertical
edge features in the image and extract the digits using a simple graphical model.
Tekim et al. [7] improvized the method in [6] by adopting a connected-component-
based approach to detect LED/LCD digits on a Nokia N95 mobile phone that can
process up to 5 frames/s allowing the user to overcome issues such as highlights,
glare, or saturation by simply varying the camera viewpoint. In [8], the authors
address a method to recognize seven-segment displays on multimeters using binary
edge and corner features. In this paper, a camera-based system is developed to detect
and recognize seven-segment displays in digital energy meters. The method can be
applied to images of several other electronic appliances such as calculators, digital
blood pressure monitors, digital thermometers, microwave ovens, media players, etc.
with minimal or no modification.

3 Proposed Method for Recognition of Seven-Segment

Digits

This section describes the proposed method designed for the recognition of seven-
segment displays from images of digital energy meters captured using mobile camera
phones. The method involves two sub-tasks, namely, (i) localization of the LCD
display area and (ii) recognition of the seven-segment digits in the localized area.
The LCD display area is localized based on attributes of horizontal and vertical line
segments and their intersection, while a support vector machine (SVM) is used for
the classification of the digits. A schematic block diagram of the proposed method
is shown in Fig. 2.

Fig. 2 Block diagram of the proposed seven-segment display recognition system

Visit https://textbookfull.com
now to explore a rich
collection of eBooks, textbook
and enjoy exciting offers!
4 T. Kasar

3.1 Determination of the LCD Display Area

Following image acquisition, the first task is to locate the LCD display area among
substantial amounts of background clutter. To this end, Canny edge detection [9] is
performed on each color channel of the captured image I . The overall edge map is
obtained by taking the union of the individual edge images as follows:

E = E R ∪ EG ∪ E B (1)

where ∪ denotes the union operation. Following the method in [10], run-length
count is performed on the resulting edge image E along the rows and columns
to obtain the horizontal and vertical lines, respectively. If the number of runs of
the edges starting at a pixel location exceeds a threshold value L, the segment is
accepted as a line. Short line segments and other spurious lines with run-lengths
less than the specified threshold value are not considered for further processing.
The threshold L decides the shortest line that can be detected by the method. This
parameter is adaptively set to a fractional proportion of the height of the image. It may
be mentioned that the performance of the method is not sensitive to the choice of this
parameter since the LCD display panel normally occupies a significant proportion of
the image area. The union of the set of validated horizontal and vertical line segments
obtained from the two directions yields a composite image I L . Based on the positions
of intersection of horizontal and vertical lines, their heights, and aspect ratios, the
rectangular-shaped LCD display area is identified. This step of identifying the region
of interest (ROI), i.e., the LCD display region, is an important processing module
that removes the background clutter and returns only the relevant display region for
further processing. The performance of this module is critical since it serves as the
input for the subsequent digit recognition task and affects the overall performance of
the system. It may be noted that the run-length method for horizontal and vertical line
detection implicitly assumes that orientation of the image is not skewed. However,
the method can tolerate a moderate skew angle of up to θ which is given by

θ = arctan(1/L) (2)

For instance, if the minimum detectable line length L is set to eight pixels, the skew
tolerance of the method is about ±7.125◦ . Thus, there is no strict requirement on the
orientation of the camera viewpoint during image capture.

3.2 Segmentation of Digits

Once the ROI is located as described above, the area defined by the ROI is cropped
off from the image. Since seven-segment LCD displays are represented in a darker
shade with respect to that of the background, the strategy is to look for thin dark
Recognition of Seven-Segment Displays from Images of Digital Energy Meters 5

structures in the detected ROI. To enhance dark and thin line-like structures, the
grayscale morphological bottom-hat operation is performed on the smoothed image
obtained by Gaussian filtering the ROI.

I p = (Iσ • S N ) − Iσ (3)

where Iσ = Ig ∗ G σ with σ representing the variance of the Gaussian filter, Ig the

grayscale image patch defined by the ROI and S N is a square structuring element of
size N × N . The notations “∗” and “•” denote the 2-D convolution and the grayscale
morphological closing operation, respectively. The Gaussian filter reduces the effect
of noise and helps to maintain the continuity of narrow gaps between line segments.
The variance σ of the Gaussian function controls the amount of smoothing. The
bottom-hat operation enhances small dark structures while suppressing wide dark
structures at the same time. The size N of the structuring element decides the max-
imum width of the line that can be detected by the system and is empirically set to
15 in this work. The method is not very sensitive to the choice of this parameter and
it may be altered without any significant effect on the overall performance.
This intermediate image I p is then thresholded using a fast implementation of
the Niblack’s method [11] using integral images [12]. At each pixel location (x, y),
the mean μ(x, y) and the standard deviation σ(x, y) within a window W × W are
computed and the gray level at that pixel location is compared with the threshold
value T (x, y) given by the following expression:

T (x, y) = μ(x, y) − kσ(x, y) (4)

The window of size is set to h/5 × w/5 where h and w denote the height and
width of the detected ROI, respectively, and the parameter k is set to 0.2. Since
the segments of a seven-segment character are not connected, we need to group
the individual segments to form the digits before feature extraction and recognition.
An eight connected component (CC) labeling is performed on the resulting binary
image. Components that touch the image boundary are discarded since it is generally
a part of the background. The stroke widths of the remaining CCs are computed
using a combination of the distance transform and the skeleton of the CC obtained
using a fast thinning algorithm proposed by Zhang and Suen [13]. The maximum
stroke width Dmax is then determined which is used to group the CCs into digits by
performing a closing operation with square structuring of size 2Dmax × 2Dmax . One
further step of CC labeling is performed on the resulting image and the bounding box
attributes are computed. Since the digits in LCD displays are of similar heights and
located horizontally next to each other, the candidate digits are obtained by imposing
the height similarity and spatial regularity of the CCs. These filtered CCs are then
passed onto the recognition module.
6 T. Kasar

Fig. 3 The seven-segment

font and the code (abcdefg)
for each digit from 0–9 in
terms of its ON/OFF states
of the individual segments

3.3 Digit Recognition

Figure 3 depicts a typical seven-segment display unit, where any digit can be repre-
sented by a 7-bit code depending on the ON or OFF state of the individual segments
a, b, ..., g. Since seven-segment displays have a fixed font style, a simple digit rec-
ognizer may be formulated based on syntactic rules to recognize the 10 digits [0 1
2 3 4 5 6 7 8 9]. However, such deterministic rules may not work well in practice
due to segmentation errors and noise. Here, in this work, an SVM classifier trained
on a collection of 500 digits is used. For each segmented CC, the proportion of ON
pixels within an area defined by each of individual segments is measured. From
seven such measurements over each of the areas defined by the seven segments, a
seven-dimensional feature vector is obtained from each candidate CC which is then
classified using the trained SVM classifier. While the aspect ratio (width/height) for
each digit is around 0.5, the same parameter is much smaller for the digit 1 and hence
it can be easily identified. Whenever the aspect ratio of the test CC is less the 0.4,
feature extraction is done only for the segments b and c and assign 0 for all the other
segments.

3.3.1 Identification of Dots

As a post-processing step, the area within the recognized digits is examined in the
presence of dots. Identification of a dot is based on the size and aspect ratio of the
CCs. Any small CCs that lie between two segmented digits are separately processed
to determine if it is a dot or not. If the size of such a CC is less than t1 times π Dmax 2

(t1 is a scalar, set to 1.5 in this work) and its aspect ratio is close to 1, it is considered
as a dot.

3.4 Experimental Results

To test the performance of the system, 175 images of energy meters are captured
using Apple iPhone 4S and Tinno S4700. These images are captured from a distance
Recognition of Seven-Segment Displays from Images of Digital Energy Meters 7

Fig. 4 Intermediate results of the proposed method on a sample test image a Detection of horizontal
and vertical line segments b Localization of the LCD display area c Segmentation of the digits d
Final results of recognition. Note that the decimal point indicated by the cyan rectangle is also
identified thereby achieving a correct recognition of the value of the reading on the device

of about 15 to 30 cms from the device and ensuring that the orientation of the captured
image is roughly horizontal. However, there is no strict requirement for the position
and orientation of the camera viewpoint during image capture. The low and high
threshold parameters of the Canny edge detection are set to 0.1 and 0.2, respectively,
while the variance of the associated Gaussian function is set to 2. The parameter L
for horizontal and vertical line segment detection is adaptively set to 1/30 times the
height of the image. Since the LCD display panel of the device is rectangular with
a fixed aspect ratio (between 3 to 4), the display area can be identified by subject-
ing each pair of intersecting horizontal and vertical lines to an aspect ratio test and
additionally a lower bound on the length of the vertical line segment to filter out
small candidate regions. Whenever there is a detection of nested regions, the overall
bounding rectangle of all the detected regions is considered to be the ROI. Following
the localization of LCD display area, a preprocessing step of Gaussian smoothing
and grayscale morphological black-hat operation is performed to enhance the seg-
ments of the seven-segment characters. While conventional thresholding techniques
fail to accurately extract the digits, the black-hat operation enables reliable digit
8 T. Kasar

2 1 9 4
6 9

0 3 0 6 1 6
1 1

6 9 8

2 8 2 0
2 3 2 3

Fig. 5 Representative results of the proposed method. The segmented digits are represented in
terms of the bounding boxes and the corresponding recognized results are overlaid above each
digit. The method can handle images exhibiting uneven illumination, the presence of shadows, poor
contrast, and blur
Random documents with unrelated
content Scribd suggests to you:
 “That’s because Father and Mother can’t get out of their grooves,
and there are so few economies which seem worth while. I believe we
need nearly fifty thousand to clear the estate.”
“But it’s silly to do nothing.”
“I don’t see what we can do. But I never could understand about
mortgages.”
“Nor could I. The only thing I can make out is that our grandfather
was a pretty awful fool.”
“He couldn’t read the future. He couldn’t tell the price of land was
going down with a bump, and that there would be a European war. I
believe we’d have been all right if it hadn’t been for the war.”
“No we shouldn’t—we were going down hill before that. The war
only hurried things on.”
“Well certainly it didn’t do for us what it did for Julian—Seventy
thousand pounds that man’s made out of blood.”
“Then I really do think he might let us have some of it. What’s
Mary’s opinion?”
Jenny shrugged.
“Oh, I dunno. He’s had a row with her too.”
“What?—about the same thing?”
“No—about a man she’s friends with. It’s ridiculous really, for he’s
years and years older than she is—a retired naval officer—and
awfully nice; I lunched with them both once in town. But it pleases
Julian to be jealous, and I believe poor Mary’s had a hideous time.”
“Lord! What upheavals since I was home last! Why doesn’t anyone
ever write and tell me about these things?”
“Because we’re all too worried and too lazy. But you’ve heard
everything now—and you really must come down and see Peter.”
“I’m coming in a moment. But tell me first—has he changed at all?
It’s more than a year since I saw him.”
“I don’t think he’s changed much, except that he’s got stouter.”
“I wonder what he’ll do with himself now he’s home. Is there really
a rumour, or have I only dreamed, that he’s keen on Stella Mount?”
“Oh, I believe he’s keen enough. But she hasn’t got a penny. Father
will be sick if he marries her.”
 She switched off the light, and the window changed from a deep,
undetailed blue to a pallid, star-pricked grey, swept across by the
tossing branches of trees.
 §5
At Conster Manor dinner was always eaten in state. Lady Alard
took hers apart in her sitting-room, and sometimes Doris had it with
her. On his “bad days” Sir John was wont to find Doris a convenient
butt, and as she was incapable either of warding off or receiving
gracefully the arrows of his wrathful wit, she preserved her dignity by
a totally unappreciated devotion to her mother. Tonight, however,
she could hardly be absent, in view of Peter’s return, and could only
hope that the presence of the heir would distract her father from his
obvious facilities.
George and Rose had stayed to dinner in honour of the occasion or
rather had come back from a visit to Leasan Vicarage for the purpose
of changing their clothes. Rose always resented having to wear
evening dress when “just dining with the family.” At the Rectory she
wore last year’s summer gown, and it seemed a wicked waste to have
to put on one of her only two dance frocks when invited to Conster.
But it was a subject on which Sir John had decided views.
“Got a cold in your chest, Rose?” he had inquired, when once she
came in her parsonage voile and fichu, and on another occasion had
coarsely remarked: “I like to see a woman’s shoulders. Why don’t you
show your shoulders, Rose? In my young days every woman showed
her shoulders if she’d got any she wasn’t ashamed of. But nowadays
the women run either to bone or muscle—so perhaps you’re right.”
Most of the Alard silver was on the table—ribbed, ponderous stuff
of eighteenth century date, later than the last of the lost causes in
which so much had been melted down. Some fine Georgian and
Queen Anne glass and a Spode dinner-service completed the
magnificence, which did not, however, extend to the dinner itself.
Good cooks were hard to find and ruinously expensive, requiring also
their acolytes; so the soup in the Spode tureen might have appeared
on the dinner-table of a seaside boarding-house, the fish was
represented by greasily fried plaice, followed by a leg of one of the
Conster lambs, reduced by the black magic of the kitchen to the
flavour and consistency of the worst New Zealand mutton.
 Peter noted that things had “gone down,” and had evidently been
down for a considerable time, judging by the placidity with which
(barring a few grumbles from Sir John) the dinner was received and
eaten. The wine, however, was good—evidently the pre-war cellar
existed. He began to wonder for the hundredth time what he had
better do to tighten the Alard finances—eating bad dinners off costly
plate seemed a poor economy. Also why were a butler and two
footmen necessary to wait on the family party? The latter were hard-
breathing young men, recently promoted from the plough, and
probably cheap enough, but why should his people keep up this
useless and shoddy state when their dear lands were in danger?
Suppose that in order to keep their footmen and their silver and their
flowers they had to sell Ellenwhorne or Glasseye—or, perhaps, even
Starvecrow....
After the dessert of apples from Conster orchard and a dish of
ancient nuts which had remained untasted and unchanged since the
last dinner-party, the women and Gervase left the table for the
drawing-room. Gervase had never sought to emphasise his man’s
estate by sitting over his wine—he always went out like this with the
women, and evidently meant to go on doing so now he had left
school. George on the other hand remained, though he rather
aggressively drank nothing but water.
“It’s not that I consider there is anything wrong in drinking wine,”
he explained broad-mindedly to Sir John and Peter, “but I feel I must
set an example.”
“To whom?” thundered Sir John.
“To my parishioners.”
“Well, then, since you’re not setting it to us, you can clear out and
join the ladies. I won’t see you sit there despising my port—which is
the only good port there’s been in the Rye division since ’16—besides
I want a private talk with Peter.”
The big clergyman rose obediently and left the room, his feelings
finding only a moment’s expression at the door, when he turned
round and tried (not very successfully) to tell Peter by a look that Sir
John must not be allowed to drink too much port in his gouty
condition.
 “He’s a fool,” said his father just before he had shut the door. “I
don’t know what the church is coming to. In my young days the
Parson drank his bottle with the best of ’em. He didn’t go about being
an example. Bah! who’s going to follow Georgie’s example?”
“Who, indeed?” said Peter, who had two separate contempts for
parsons and his brother George, now strengthened by combination.
“Well, pass me the port anyhow. Look here, I want to talk to you—
first time I’ve got you alone. What are you going to do now you’re
back?”
“I don’t know, Sir. I’ve scarcely had time to think.”
“You’re the heir now, remember. I’d rather you stayed here. You
weren’t thinking of going back into Lightfoot’s, were you?”
“I don’t see myself in the city again. Anyhow I’d sooner be at
Conster.”
“That’s right. That’s your place now. How would you like to be
Agent?”
“I’d like it very much, Sir. But can it be done? What about
Greening?”
“He’s an old fool, and has been muddling things badly the last year
or two. He doesn’t want to stay. I’ve been talking to him about
putting you in, and he seemed glad.”
“I’d be glad too, Sir.”
“You ought to know more about the estate than you do. It’ll be
yours before long—I’m seventy-five, you know. When Hugh was alive
I thought perhaps a business career was best for you, so kept you out
of things. You’ll have to learn a lot.”
“I love the place, Sir—I’m dead keen.”
“Yes, I remember you always wanted.... Of course I’ll put you into
Starvecrow.”
“Starvecrow!”
“Don’t repeat my words. The Agent has always lived at Starvecrow,
and there are quite enough of us here in the house. Besides there’s
another thing. How old are you?”
“Thirty-six.”
“Time you married, ain’t it?”
 “I suppose it is.”
“I was thirty, myself, when I married, but thirty-six is rather late.
How is it you haven’t married earlier?”
“Oh, I dunno—the war I suppose.”
“The war seems to have had the opposite effect on most people.
But my children don’t seem a marrying lot. Doris ... Hugh ... there’s
Mary, of course, and George, but I don’t congratulate either of ’em.
Julian’s a mean blackguard, and Rose——” Sir John defined Rose in
terms most unsuitable to a clergyman’s wife.
“You really must think about it now,” he continued—“you’re the
heir; and of course you know—we want money.”
Peter did not speak.
“We want money abominably,” said Sir John, “in fact I don’t know
how we’re to carry on much longer without it. I don’t want to have to
sell land—indeed, it’s practically impossible, all trussed up as we are.
Starvecrow could go, of course, but it’s useful for grazing and
timber.”
“You’re not thinking of selling Starvecrow?”
“I don’t want to—we’ve had it nearly two hundred years; it was the
first farm that Giles Alard bought. But it’s also the only farm we’ve
got in this district that isn’t tied—there’s a mortgage on the grazings
down by the stream, but the house is free, with seventy acres.”
“It would be a shame to let it go.”
Peter was digging into the salt-cellar with his dessert knife.
“Well, I rely on you to help me keep it. Manage the estate well and
marry money.”
“You’re damn cynical, Sir. Got any especial—er—money in your
mind?”
“No, no—of course not. But you ought to get married at your age,
and you might as well marry for the family’s advantage as well as
your own.”
Peter was silent.
“Oh, I know there’s a lot to be said against getting married, but in
your position—heir to a title and a big estate—it’s really a duty. I
married directly my father died. But don’t you wait for that—you’re
getting on.”
“But who am I to marry? There’s not such a lot of rich girls round
here.”
“You’ll soon find one if you make up your mind to it. My plan is
first make up your mind to get married and then look for the girl—
not the other way round, which is what most men do, and leads to all
kinds of trouble. Of course I know it isn’t always convenient. But
what’s your special objection? Any entanglement? Don’t be afraid to
tell me. I know there’s often a little woman in the way.”
Peter squirmed at his father’s Victorian ideas of dissipation with
their “little women.” He’d be talking of “French dancers” next....
“I haven’t any entanglement, Sir.”
“Then you take my words to heart. I don’t ask you to marry for
money, but marry where money is, as Shakespeare or somebody
said.”
 §6
Peter found a refreshing solitude in the early hours of the next day.
His mother and Doris breakfasted upstairs, his father had
characteristically kept his promise to “be about tomorrow,” and had
actually ridden out before Peter appeared in the morning room at
nine. Jenny, who was a lazy young woman, did not come down till he
had finished, and Gervase, in one of those spasms of eccentricity
which made Peter sometimes a little ashamed of him, had gone
without breakfast altogether, and driven off in the Ford lorry to fetch
his luggage, sustained by an apple.
The morning room was full of early sunlight—dim as yet, for the
mists were still rising from the Tillingham valley and shredding
slowly into the sky. The woods and farms beyond the river were
hidden in the same soft cloud. Peter opened the window, and felt the
December rasp in the air. Oh, it was good to be back in this place,
and one with it now, the heir.... No longer the second son who must
live away from home and make his money in business.... He stifled
the disloyalty to his dead brother. Poor old Hugh, who was so solemn
and so solid and so upright.... But Hugh had never loved the place as
he did—he had never been both transported and abased by his
honour of inheritance.
As soon as he had eaten his breakfast Peter went out, at his heels a
small brown spaniel, who for some reason had not gone with the
other dogs after Sir John. They went down the garden, over the half
melted frost of the sloping lawns, through the untidy shrubbery of
fir, larch and laurel, to the wooden fence that shut off Conster from
the marshes of the Tillingham. The river here had none of the
pretensions with which it circled Rye, but was little more than a
meadow-stream, rather full and angry with winter. Beyond it, just
before the woods began, lay Beckley Furnace with its idle mill.
And away against the woods lay Starvecrow ... just as he had
dreamed of it so many times in France, among the blasted fields.
“Starvecrow”—he found himself repeating the name aloud, but not as
it was written on the map, rather as it was written on the lips of the
people to whom its spirit belong—“Starvycrow ... Starvycrow.”
 It was a stone house built about the same time as Conster, but
without the compliment to Gloriana implied in three gables. It lacked
the grace of Conster—the grace both of its building and of its
planting. It stood foursquare and forthright upon the slope, with a
great descent of wavy, red-brown roof towards the mouth of the
valley, a shelter from the winds that came up the Tillingham from the
sea. It seemed preeminently a home, sheltered, secure, with a
multitude of chimneys standing out against the background of the
woods. From one of them rose a straight column of blue smoke,
unwavering in the still, frost-thickened air.
Peter crossed the stream by the bridge, then turned up
Starvecrow’s ancient drive. There was no garden, merely an orchard
with a planting of flowers under the windows. Peter did not ring, but
walked straight in at the side door. The estate office had for long
years been at Starvecrow, a low farmhouse room in which the office
furniture looked incongruous and upstart.
“I’ll change all this,” thought Peter to himself—I’ll have a gate-
legged table and Jacobean chairs.
The room was empty, but the agents wife had heard him come in.
“That you, Mr. Alard? I thought you’d be over. Mr. Greening’s gone
to Winterland this morning. They were complaining about their roof.
He said he’d be back before lunch.”
Peter shook hands with Mrs. Greening and received rather
abstractedly her congratulations on his return. He was wondering if
she knew he was to supplant them at Starvecrow.
She did, for she referred to it the next minute, and to his relief did
not seem to resent the change.
“We’re getting old people, and for some time I’ve been wanting to
move into the town. It’ll be a good thing to have you here, Mr. Alard
—bring all the tenants more in touch with the family. Not that Sir
John doesn’t do a really amazing amount of work....”
She rambled on, then suddenly apologised for having to leave him
—a grandchild staying in the house was ill.
“Shall you wait for Mr. Greening? I’m afraid he won’t be in for an
hour at least.”
“I’ll wait for a bit anyway. I’ve some letters to write.”
 He went into the office and sat down. The big ugly rolltop desk was
littered with papers—memoranda, bills, estimates, plans of farms,
lists of stock-prices. He cleared a space, seized a couple of sheets of
the estate note-paper, and began to write.
“My loveliest Stella,” he wrote.
 §7
He had nearly covered the two sheets when the rattle of a car
sounded in the drive below. He looked up eagerly and went to the
window, but it was only Gervase lurching over the ruts in the Ford,
just scraping past the wall as he swung round outside the house, just
avoiding a collision with an outstanding poplar, after the usual
manner of his driving.
The next minute he was in the office.
“Hullo! They told me you were over here. I’ve just fetched my
luggage from Robertsbridge.”
He sat down on the writing-table and lit a cigarette. Peter hastily
covered up his letter. Why did Gervase come bothering him now?
“I wanted to speak to you,” continued his brother. “You’ll be the
best one to back me up against Father.”
“What is it now?” asked Peter discouragingly.
“An idea came to me while I was driving over. I often get ideas
when I drive, and this struck me as rather a good one. I think it
would be just waste for me to go to a crammer’s and then to Oxford. I
don’t want to go in for the church or the bar or schoolmastering or
anything like that, and I don’t see why the family should drop
thousands on my education just because I happen to be an Alard. I
want to go in for engineering in some way and you don’t need any
’Varsity for that. I could go into some sort of a shop....”
“Well, if the way you drive a car is any indication——”
“I can drive perfectly well when I think about it. Besides, that
won’t be my job. I want to learn something in the way of construction
and all that. I always was keen, and it strikes me now that I’d much
better go in for that sort of thing than something which won’t pay for
years. There may be some sort of a premium to fork out, but it’ll be
nothing compared to what it would cost to send me to Oxford.”
“You talk as if we were paupers,” growled Peter.
“Well, so we are, aren’t we?” said Gervase brightly. “Jenny was
talking to me about it last night. She says we pay thousands a year in
interest on mortgages, and as for paying them off and selling the
land, which is the only thing that can help us....”
“I don’t see that it’s your job, anyway.”
“But I could help. Really it seems a silly waste to send me to
Oxford when I don’t want to go.”
“You need Oxford more than any man I know. If you went there
you might pick up some notions of what’s done, and get more like
other people.”
“I shouldn’t get more like other people, only more like other
Oxford men.”
Peter scowled. He intensely disapproved of the kid’s verbal
nimbleness, which his more weighty, more reputable argument could
only lumber after.
“You’ve got to remember you’re a gentleman’s son,” he remarked
in a voice which suggested sitting down just as Gervase’s had
suggested a skip and a jump.
“Well, lots of them go in for engineering. We’re in such a groove. I
daresay you think this is just a sudden idea of mine——”
“You’ve just told me it is.”
“I know, but I’ve been thinking for ages that I didn’t want to go to
Oxford. If I took up engineering I could go into a shop at Ashford....
But I’ll have to talk to Father about it. I expect he’ll be frightfully
upset—the only Alard who hasn’t been to the Varsity and all that ...
but, on the other hand, he’s never bothered about me so much as
about you and George, because there’s no chance of my coming into
the estate.”
“I wouldn’t be too sure,” gibed Peter.
“Yes, of course, you might both die just to spite me—but it
wouldn’t be sporting of you. I don’t want to be Sir Gervase Alard,
Bart.—I’d much rather be Alard and Co., Motor-engineers.”
“You damn well shan’t be that.”
“Well, it’s a long time ahead, anyway. But do back me up against
Father about not going to Oxford. It really ought to help us a lot if I
don’t go—a son at the ’Varsity’s a dreadful expense, and when that
son’s me, it’s a waste into the bargain.”
 “Well, I’ll see about it. My idea is that you need Oxford more than
—hullo, who’s that?”
“Dr. Mount,” said Gervase looking out of the window.
Peter rose and looked out too, in time to see the doctor’s car
turning in the sweep. This morning he himself was not at the wheel,
but was driven by what looked like a warm bundle of furs with a pair
of bright eyes looking out between collar and cap.
Peter opened the window.
“Stella!” he cried.
 §8
A minute later Stella Mount was in the room. Gervase had not seen
her for several years; during the greater part of the war she had been
away from home, first at a munition factory, then as an auxiliary
driver to the Army Service Corps. When last they had met the gulf
between the schoolboy of fourteen and the girl of twenty had yawned
much wider than between the youth of eighteen and the young
woman of twenty-four. Stella looked, if anything, younger than she
had looked four years ago, and he was also of an age to appreciate
her beauty which he had scarcely noticed on the earlier occasions.
In strict point of fact Stella was not so much beautiful as pretty, for
there was nothing classic in her little heart-shaped face, with its wide
cheek-bones, pointed chin and puckish nose. On the other hand
there was nothing of that fragile, conventional quality which
prettiness is understood to mean. Everything about Stella was
healthy, warm and living—her plump little figure, the glow on her
cheeks, the shine of her grey eyes between their lashes, like pools
among reeds, the decision of her chin and brows, the glossy,
tumbling masses of her hair, all spoke of strength and vigour, a
health that was almost hardy.
She came into the room like a flame, and Gervase felt his heart
warming. Then he remembered that she was Peter’s—Jenny had said
so, though she had not blessed Peter’s possession.
“How d’you do, Stella?” he said, “it’s ages since we met. Do you
know who I am?”
“Of course I do. You haven’t altered much, except in height. You’ve
left Winchester for good now, haven’t you?”
“Yes—and I’ve just been arguing with Peter about what I’m to do
with myself now I’m home.”
“How very practical of you! I hope Peter was helpful.”
“Not in the least.”
He could feel Peter’s eyes upon him, telling him to get out of the
way and leave him alone with his bright flame....
 “Well, I must push off—they may be wanting the Ford at home.”
He shook hands with Stella, nodded to Peter, and went out.
For a moment Peter and Stella faced each other in silence. Then
Peter came slowly up to her and took her in his arms, hiding his face
in her neck.
“O Stella—O my beauty!...”
She did not speak, but her arms crept round him. They could
scarcely meet behind his broad back—she loved this feeling of girth
which she could not compass, combined as it was with a queer tender
sense of his helplessness, of his dependence on her——
“O Peter,” she whispered—“my little Peter....”
“I was writing to you, darling, when you came.”
“And I was on my way to see you at Conster. Father was going
there after he’d called on little Joey Greening. I wouldn’t come
yesterday—I thought your family would be all over you, and I didn’t
like....”
She broke off the sentence and he made no effort to trim the
ragged end. Her reference to his family brought back into his
thoughts the conversation he had had with his father over the wine.
She had always felt his family as a cloud, as a barrier between them,
and it would be difficult to tell her that now he was the heir, now he
was home from the war, instead of being removed the cloud would be
heavier and the barrier stronger.
“I’m so glad you came here”—he breathed into her hair—“that our
first meeting’s at Starvecrow.”
“Yes—I’m glad, too.”
Peter sat down in the leather-covered office chair, holding Stella
on his knee.
“Child—they’re going to give me Starvecrow.”
“O Peter!”...
“Yes—Greening wants to leave, and my father’s making me agent
in his place.”
“How lovely! Shall you come and live here?”
“Yes.”
 The monosyllable came gruffly because of the much more that he
wanted to say. It was a shame to have such reserves spoil their first
meeting.
“I’m so awfully, wonderfully glad, Peter darling.”
She hid her soft, glowing face in his neck—she was lying on his
breast like a child, but deliciously heavy, her feet swung off the floor.
“Stella—my sweetheart—beautiful....”
His love for her gave him a sweet wildness of heart, and he who
was usually slow of tongue, became almost voluble——
“Oh, I’ve longed for this—I’ve thought of this, dreamed of this....
And you’re lovelier than ever, you dear.... Stella, sweetheart, let me
look into your eyes—close to—like that ... your eyelashes turn back
like the petals of a flower.... O you wonderful, beautiful thing ... And
it’s so lovely we should have met here instead of at home—the
dearest person in the dearest place ... Stella at Starvecrow.”
“Starvycrow,” she repeated gently.
For a moment he felt almost angry that she should have used his
name—his private music. But his anger melted into his love. She
used his name because she, alone in all the world, felt his feelings
and thought his thoughts. Perhaps she did not love Starvecrow quite
as he did, but she must love it very nearly as much or she would not
call it by its secret name. They sat in silence, her head upon his
shoulder, his arms about her, gathering her up on his knees. On the
hearth a log fire softly hummed and sighed. Ages seemed to flow over
them, the swift eternities of love.... Then suddenly a voice called
“Stella!” from the drive.
She started up, and the next moment was on her feet, pushing
away her hair under her cap, buttoning her high collar over her chin.
“How quick Father’s been! I feel as if I’d only just come.”
“You must come again.”
“I’m coming to dinner on Christmas day, you know.”
“That doesn’t count. I want you here.”
“And I want to be here with you—always.”
The last word was murmured against his lips as he kissed her at
the door. He was not quite sure if he had heard it. During the rest of
the morning he sometimes feared not—sometimes hoped not.
 §9
“It will be a green Christmas,” said Dr. Mount.
Stella made no answer. The little car sped through the lanes at the
back of Benenden. They had driven far—to the very edge of the
doctor’s wide-flung practice, by Hawkhurst and Skullsgate, beyond
the Kent Ditch. They had called at both the Nineveh farms—Great
Nineveh and Little Nineveh—and had now turned south again. The
delicate blue sky was drifted over with low pinkish clouds, which
seemed to sail very close to the field where their shadows moved; the
shadows swooped down the lane with the little car, rushing before it
into Sussex. Stella loved racing the sky.
On her face, on her neck, she could still feel cold places where
Peter had kissed her. It was wonderful and beautiful, she thought,
that she should carry the ghosts of his kisses through Sussex and
Kent. And now she would not have so long to be content with ghosts
—there would not be those terrible intervals of separation. She would
see Peter again soon, and the time would come—must come—when
they would be together always. “Together always” was the fulfilment
of Stella’s dream. “They married and were together always” sounded
better in her ears than “they married and lived happy ever after.” No
more partings, no more ghosts of kisses, much as she loved those
ghosts, but always the dear, warm bodily presence—Peter working,
Peter resting, Peter sleepy, Peter hungry, Peter talking, Peter silent—
Peter always.
“It will be a green Christmas,” repeated Dr. Mount.
“Er—did you speak, Father dear?”
“Yes, I said it would be a gr——but never mind, I’m sure your
thoughts are more interesting than anything I could say.”
Stella blushed. She and her father had a convention of silence
between them in regard to Peter. He knew all about him, of course,
but they both pretended that he didn’t; because Stella felt she had no
right to tell him until Peter had definitely asked her to be his wife.
And he had not asked her yet. When they had first fallen in love,
Hugh Alard was still alive and the second son’s prospects were
uncertain; then when Hugh was killed and Peter became the heir,
there was still the war, and she knew that her stolid, Saxon Peter
disapproved of war-weddings and grass widows who so often became
widows indeed. He had told her then he could not marry her till after
the war, and she had treated that negative statement as the
beginning of troth between them. She had never questioned or
pressed him—it was not her way—she had simply taken him for
granted. She had felt that he could not, any more than she, be
satisfied with less than “together always.”
But now she felt that something definite must happen soon, and
their tacit understanding become open and glorious. His family
would disapprove, she knew, though they liked her personally and
owed a great deal to her father. But Stella, outside and unaware,
made light of Conster’s opposition. Peter was thirty-six and had five
hundred a year of his own, so in her opinion could afford to snap his
fingers at Alard tyranny. Besides, she felt sure the family would
“come round”—they would be disappointed at first, but naturally
they wouldn’t expect Peter to give up his love-choice simply because
she had no money. She would be glad when things were open and
acknowledged, for though her secret was a very dear one, she was
sometimes worried by her own shifts to keep it, and hurt by Peter’s.
It hurt her that he should have to pretend not to care about her when
they met in public—but not so much as it would have hurt her if he
hadn’t done it so badly.
“Well, now he’s back, I suppose Peter will take the eldest son’s
place,” said Dr. Mount, “and help his father manage the property.”
“Yes—he told me this morning that Sir John wants him to be agent
instead of Mr. Greening, and he’s to live at Starvecrow.”
“At Starvecrow! You’ll like that—I mean, it’s nice to think Peter
won’t have to go back and work in London. I always felt he belonged
here more than Hugh.”
“Yes, I don’t think Hugh cared for the place very much, but Peter
always did. It always seemed hard lines that he should be the second
son.”
“Poor Hugh,” said Dr. Mount—“he was very like Peter in many
ways—Sober and solid and kind-hearted; but he hadn’t Peter’s
imagination.”
“Peter’s very sensitive,” said Stella—“in spite of his being such a
big, heavy thing.”
Then she smiled, and said in her heart—“Peter’s mine.”
 § 10
Christmas was celebrated at Conster in the manner peculiar to
houses where there is no religion and no child. Tradition compelled
the various members of the family to give each other presents which
they did not want and to eat more food than was good for them; it
also compelled them to pack unwillingly into the Wolsey car and
drive to Leasan church, where they listened in quite comprehensible
boredom to a sermon by brother George. Peter was able to break free
from this last superstition, and took himself off to the office at
Starvecrow—his family’s vague feeling of unrest at his defection
being compensated by the thought that there really wouldn’t have
been room for him in the car.
But Starvecrow was dim and sodden on this green Christmas day,
full of a muggy cloud drifting up from the Tillingham, and Peter was
still sore from the amenities of the Christmas breakfast table—that
ghastly effort to be festive because it was Christmas morning, that
farce of exchanging presents—all those empty rites of a lost
childhood and a lost faith. He hated Christmas.
Also he wanted Stella, and she was not to be had. She too had gone
to church—which he would not have minded, if she had not had the
alternative of being with him here at Starvecrow. He did not at all
object to religion in women as long as they kept it in its proper place.
But Stella did not keep hers in its proper place—she let it interfere
with her daily life—with his ... and she had not gone to church at
Leasan, which was sanctified to Peter by the family patronage and
the family vault, but to Vinehall, where they did not even have the
decencies of Dearly Beloved Brethren, but embarrassing mysteries
which he felt instinctively to be childish and in bad taste.
In Stella’s home this Christmas there would be both religion and
children, the latter being represented by her father and herself. Last
night when he called at Hollingrove—Dr. Mount’s cottage on the
road between Leasan and Vinehall—to ask her to meet him here
today at Starvecrow, he had found her decorating a Christmas tree,
to be put in the church, of all places. She had asked him to stop and