Download the Full Version of textbook for Fast Typing at textbookfull.

com

Sensors and Image Processing Proceedings of CSI

2015 1st Edition Shabana Urooj

https://textbookfull.com/product/sensors-and-image-
processing-proceedings-of-csi-2015-1st-edition-shabana-
urooj/

OR CLICK BUTTON

DOWNLOAD NOW

Download More textbook Instantly Today - Get Yours Now at textbookfull.com

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

Speech and Language Processing for Human Machine

Communications Proceedings of CSI 2015 1st Edition S. S.
Agrawal
https://textbookfull.com/product/speech-and-language-processing-for-
human-machine-communications-proceedings-of-csi-2015-1st-edition-s-s-
agrawal/
textboxfull.com

Silicon Photonics amp High Performance Computing

Proceedings of CSI 2015 1st Edition Anurag Mishra

https://textbookfull.com/product/silicon-photonics-amp-high-
performance-computing-proceedings-of-csi-2015-1st-edition-anurag-
mishra/
textboxfull.com

Computational Methods and Clinical Applications for Spine

Imaging Third International Workshop and Challenge CSI
2015 Held in Conjunction with MICCAI 2015 Munich Germany
October 5 2015 Proceedings 1st Edition Tomaž Vrtovec
https://textbookfull.com/product/computational-methods-and-clinical-
applications-for-spine-imaging-third-international-workshop-and-
challenge-csi-2015-held-in-conjunction-with-miccai-2015-munich-
germany-october-5-2015-proceedings-1st/
textboxfull.com

Applied Computer Vision and Image Processing Proceedings

of ICCET 2020 Volume 1 Brijesh Iyer

https://textbookfull.com/product/applied-computer-vision-and-image-
processing-proceedings-of-iccet-2020-volume-1-brijesh-iyer/

textboxfull.com
Proceedings of the 2015 International Conference on
Communications Signal Processing and Systems 1st Edition
Qilian Liang
https://textbookfull.com/product/proceedings-of-
the-2015-international-conference-on-communications-signal-processing-
and-systems-1st-edition-qilian-liang/
textboxfull.com

Frontiers in Materials Processing Applications Research

and Technology Select Proceedings of FiMPART 2015 1st
Edition M. Muruganant
https://textbookfull.com/product/frontiers-in-materials-processing-
applications-research-and-technology-select-proceedings-of-
fimpart-2015-1st-edition-m-muruganant/
textboxfull.com

Image Operators: Image Processing in Python 1st Edition

Jason M. Kinser

https://textbookfull.com/product/image-operators-image-processing-in-
python-1st-edition-jason-m-kinser/

textboxfull.com

Advances in Image and Graphics Technologies 10th Chinese

Conference IGTA 2015 Beijing China June 19 20 2015
Proceedings 1st Edition Tieniu Tan
https://textbookfull.com/product/advances-in-image-and-graphics-
technologies-10th-chinese-conference-igta-2015-beijing-china-
june-19-20-2015-proceedings-1st-edition-tieniu-tan/
textboxfull.com

Image operators image processing in Python First Edition

Kinser

https://textbookfull.com/product/image-operators-image-processing-in-
python-first-edition-kinser/

textboxfull.com
Advances in Intelligent Systems and Computing 651

Shabana Urooj
Jitendra Virmani Editors

Sensors
and Image
Processing
Proceedings of CSI 2015
Advances in Intelligent Systems and Computing

Volume 651

Series editor
Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland
e-mail: [email protected]
About this Series
The series “Advances in Intelligent Systems and Computing” contains publications on theory,
applications, and design methods of Intelligent Systems and Intelligent Computing. Virtually
all disciplines such as engineering, natural sciences, computer and information science, ICT,
economics, business, e-commerce, environment, healthcare, life science are covered. The list
of topics spans all the areas of modern intelligent systems and computing.
The publications within “Advances in Intelligent Systems and Computing” are primarily
textbooks and proceedings of important conferences, symposia and congresses. They cover
significant recent developments in the field, both of a foundational and applicable character.
An important characteristic feature of the series is the short publication time and world-wide
distribution. This permits a rapid and broad dissemination of research results.

Advisory Board
Chairman
Nikhil R. Pal, Indian Statistical Institute, Kolkata, India
e-mail: [email protected]
Members
Rafael Bello Perez, Universidad Central “Marta Abreu” de Las Villas, Santa Clara, Cuba
e-mail: [email protected]
Emilio S. Corchado, University of Salamanca, Salamanca, Spain
e-mail: [email protected]
Hani Hagras, University of Essex, Colchester, UK
e-mail: [email protected]
László T. Kóczy, Széchenyi István University, Győr, Hungary
e-mail: [email protected]
Vladik Kreinovich, University of Texas at El Paso, El Paso, USA
e-mail: [email protected]
Chin-Teng Lin, National Chiao Tung University, Hsinchu, Taiwan
e-mail: [email protected]
Jie Lu, University of Technology, Sydney, Australia
e-mail: [email protected]
Patricia Melin, Tijuana Institute of Technology, Tijuana, Mexico
e-mail: [email protected]
Nadia Nedjah, State University of Rio de Janeiro, Rio de Janeiro, Brazil
e-mail: [email protected]
Ngoc Thanh Nguyen, Wroclaw University of Technology, Wroclaw, Poland
e-mail: [email protected]
Jun Wang, The Chinese University of Hong Kong, Shatin, Hong Kong
e-mail: [email protected]

More information about this series at http://www.springer.com/series/11156

Shabana Urooj Jitendra Virmani
•

Editors

Sensors and Image

Processing
Proceedings of CSI 2015

123
Editors
Shabana Urooj Jitendra Virmani
Gautam Buddha University Department of Electrical and
Greater Noida, Uttar Pradesh Instrumentation Engineering
India Thapar University
Patiala, Punjab
India

ISSN 2194-5357 ISSN 2194-5365 (electronic)

Advances in Intelligent Systems and Computing
ISBN 978-981-10-6613-9 ISBN 978-981-10-6614-6 (eBook)
https://doi.org/10.1007/978-981-10-6614-6
Library of Congress Control Number: 2017952512

© Springer Nature Singapore Pte Ltd. 2018

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.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

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

The last decade has witnessed remarkable changes in IT industry, virtually in all
domains. The 50th Annual Convention, CSI-2015, on the theme “Digital Life” was
organized as a part of CSI-2015, by CSI at Delhi, the national capital of the country,
during December 02–05, 2015. Its concept was formed with an objective to keep
ICT community abreast of emerging paradigms in the areas of computing tech-
nologies and more importantly looking at its impact on the society.
Information and Communication Technology (ICT) comprises of three main
components: infrastructure, services, and product. These components include the
Internet, infrastructure-based/infrastructure-less wireless networks, mobile termi-
nals, and other communication mediums. ICT is gaining popularity due to rapid
growth in communication capabilities for real-time-based applications. “Nature
Inspired Computing” is aimed at highlighting practical aspects of computational
intelligence including robotics support for artificial immune systems. CSI-2015
attracted over 1500 papers from researchers and practitioners from academia,
industry, and government agencies, from all over the world, thereby making the job
of the Programme Committee extremely difficult. After a series of tough review
exercises by a team of over 700 experts, 565 papers were accepted for presentation
in CSI-2015 during the 3 days of the convention under ten parallel tracks. The
Programme Committee, in consultation with Springer, the world’s largest publisher
of scientific documents, decided to publish the proceedings of the presented papers,
after the convention, in ten topical volumes, under ASIC series of the Springer, as
detailed hereunder:
1. Volume # 1:
ICT Based Innovations
2. Volume # 2:
Next Generation Networks
3. Volume # 3:
Nature Inspired Computing
4. Volume # 4:
Speech and Language Processing for Human-Machine
Communications
5. Volume # 5: Sensors and Image Processing
6. Volume # 6: Big Data Analytics

v
vi Preface

7. Volume # 7: Systems and Architecture

8. Volume # 8: Cyber Security
9. Volume # 9: Software Engineering
10. Volume # 10: Silicon Photonics and High Performance Computing
We are pleased to present before you the proceedings of Volume # 5 on “Sensors
and Image Processing.” The title “Sensors and Image Processing” highlights the
different applications in the field of virtual reality. It also delves into the matter as to
how robotics can be applied to strengthen modeling. The title also showcases the
various augments of latest mobile technologies, solid modeling, etc.
Sensors are used in everyday objects such as touch-sensitive elevator buttons
and lamps which dim or brighten by touching the base, besides innumerable
applications of which most people are never aware of. With advances in micro-
machinery and easy-to-use microcontroller platforms, the uses of sensors have
expanded beyond the more traditional fields of temperature, pressure, or flow
measurement, for example MARG sensors. Image processing is processing of
images using mathematical operations by using any form of signal processing for
which the input is an image, such as a photograph or video frame. The output of
image processing may be either an image or a set of characteristics or parameters
related to the image. Most image-processing techniques involve treating the image
as a two-dimensional signal and applying standard signal-processing techniques to
it. The title “Sensors and Image Processing” also amalgamates and showcases the
applications of above technologies in different research and real-time domains. The
volume includes scientific, original, and high-quality papers presenting novel
research, ideas, and explorations of new vistas in speech and language processing
such as speech recognition, text recognition, embedded platform for information
retrieval, segmentation, filtering and classification of data, emotion recognition. The
aim of this volume is to provide a stimulating forum for sharing knowledge and
results in the model, methodology, and implementations of speech and language
processing tools. Its authors are researchers and experts of these domains. This
volume is designed to bring together researchers and practitioners from academia
and industry to focus on extending the understanding and establishing new col-
laborations in these areas. It is the outcome of the hard work of the editorial team,
who have relentlessly worked with the authors and steered up the same to compile
this volume. It will be a useful source of reference for the future researchers in this
domain. Under the CSI-2015 umbrella, we received over 200 papers for this vol-
ume, out of which 29 papers are being published, after rigorous review processes,
carried out in multiple cycles.
On behalf of organizing team, it is a matter of great pleasure that CSI-2015 has
received an overwhelming response from various professionals across the country.
The organizers of CSI-2015 are thankful to the members of Advisory Committee,
Programme Committee, and Organizing Committee for their all-round guidance,
encouragement, and continuous support. We express our sincere gratitude to the
learned Keynote Speakers for support and help extended to make this event a grand
success. Our sincere thanks are also due to our Review Committee Members and the
Preface vii

Editorial Board for their untiring efforts in reviewing the manuscripts, giving
suggestions and valuable inputs for shaping this volume. We hope that all the
participated delegates will be benefitted academically and wish them for their future
endeavors.
We also take the opportunity to thank the entire team from Springer, who have
worked tirelessly and made the publication of the volume a reality. Last but not
least, we thank the team from Bharati Vidyapeeth’s Institute of Computer
Applications and Management (BVICAM), New Delhi, for their untiring support,
without which the compilation of this huge volume would not have been possible.

Greater Noida, India Shabana Urooj

Patiala, India Jitendra Virmani
March 2017
The Organization of CSI-2015

Chief Patron
Padmashree Dr. R. Chidambaram, Principal Scientific Advisor, Government of
India

Patrons
Prof. S.V. Raghavan, Department of Computer Science, IIT Madras, Chennai
Prof. Ashutosh Sharma, Secretary, Department of Science and Technology,
Ministry of Science and Technology, Government of India
Chair, Programme Committee
Prof. K.K. Aggarwal, Founder Vice Chancellor, GGSIP University, New Delhi
Secretary, Programme Committee
Prof. M.N. Hoda, Director, Bharati Vidyapeeth’s Institute of Computer
Applications and Management (BVICAM), New Delhi

Advisory Committee
• Padma Bhushan Dr. F.C. Kohli, Co-Founder, TCS
• Mr. Ravindra Nath, CMD, National Small Industries Corporation, New Delhi
• Dr. Omkar Rai, Director General, Software Technological Parks of India (STPI),
New Delhi
• Adv. Pavan Duggal, Noted Cyber Law Advocate, Supreme Courts of India
• Prof. Bipin Mehta, President, CSI
• Prof. Anirban Basu, Vice President—cum- President Elect, CSI
• Shri Sanjay Mohapatra, Secretary, CSI
• Prof. Yogesh Singh, Vice Chancellor, Delhi Technological University, Delhi
• Prof. S.K. Gupta, Department of Computer Science and Engineering, IIT, Delhi

ix
x The Organization of CSI-2015

• Prof. P.B. Sharma, Founder Vice Chancellor, Delhi Technological University,

Delhi
• Mr. Prakash Kumar, IAS, Chief Executive Officer, Goods and Services Tax
Network (GSTN)
• Mr. R.S. Mani, Group Head, National Knowledge Networks (NKN), NIC,
Government of India, New Delhi

Editorial Board
• M.U. Bokhari, AMU, Aligarh
• D.K. Lobiyal, JNU, New Delhi
• Umang Singh, ITS, Ghaziabad
• Shiv Kumar, CSI
• Vishal Jain, BVICAM, New Delhi
• Shalini Singh Jaspal, BVICAM, New Delhi
• S.S. Agrawal, KIIT, Gurgaon
• Amita Dev, BPIBS, New Delhi
• Aasim Zafar, AMU, Aligarh
• Ritika Wason, BVICAM, New Delhi
• Anupam Baliyan, BVICAM, New Delhi
• S.M.K. Quadri, JMI, New Delhi
Contents

A Comparative Study of Various Color Texture Features for Skin

Cancer Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Maurya Ritesh and Srivastava Ashwani
Pattern Classification and Retrieval of Content-Based Images—a Step
Towards Amelioration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Hemjot and Amitabh Sharma
Novel Threshold Formulation for Energy Detection Method to
Efficient Spectrum Sensing in Cognitive Radio . . . . . . . . . . . . . . . . . . . . . 25
Rohini S. Kale, Vijay M. Wadhai and Jagdish B. Helonde
A Novel Architecture to Crawl Images Using OAI-PMH. . . . . . . . . . . . . 37
Shruti Sharma, Parul Gupta and C.K. Nagpal
A Novel Web Page Change Detection Technique for Migrating
Crawlers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Ashlesha Gupta, Ashutosh Dixit and A.K. Sharma
Developing Prototype for Prosopagnosia Using PCA . . . . . . . . . . . . . . . . 59
Gunjan Jhawar, Prajacta Nagraj and N. Ramesh Babu
Implementation of Smart Video Surveillance System Using Motion
Detection Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Gaurav Verma, Swati Gautam, Rohan Agarwal, Sarthak Saxena
and Deval Verma
Intelligent Algorithm for Automatic Multistoried Parking System
Using Image Processing with Vehicle Tracking and Monitoring from
Different Locations in the Building . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Sagar Juneja, Saurav Kochar and Sachin Dhiman
Computer-Aided Detection of Breast Cancer Using Pseudo Zernike
Moment as Texture Descriptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Shabana Urooj, Satya P. Singh and A.Q. Ansari

xi
xii Contents

Analysis of Leukoderma Images Using Neuro-Fuzzy Hybrid

Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Sudhakar Singh, Shabana Urooj and Satya P. Singh
Texture and Clustering-based Skin Disease Classification . . . . . . . . . . . . 103
Pradeep Mullangi, Y. Srinivasa Rao and Pushpa Kotipalli
Melanoma Skin Cancer Detection Using Image Processing . . . . . . . . . . . 111
Nishtha Garg, Vishakha Sharma and Prabhjot Kaur
Performance Assessment for Clustering Techniques for Image
Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Jyoti Arora, Kiran Khatter and Meena Tushir
Automatic Seed Point Selection in B-Mode Breast
Ultrasound Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Madan Lal and Lakhwinder Kaur
EEA-LEACH—Trustworthy-Enhanced Algorithm for Energy
Optimization in Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . . . . . . 139
Umang, M.N. Hoda and Geetanjali Singh
Secure Cluster-Based Data Integrity Routing for Wireless Sensor
Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Shashikala and C. Kavitha
Objects Detection and Tracking Strategies . . . . . . . . . . . . . . . . . . . . . . . . 169
Sureshbabu Matla and Ravi Subban
Road Accident Informer System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
S. Vanitha
Hybrid Intelligent Algorithm for Energy-Efficient Routing in WSN . . . . 189
Raminder Singh Uppal
A Method for Secret Image Transmission to Preserve Privacy . . . . . . . . 201
Rasika Thakare, B Sumit and Umesh Kulkarni
Text Retrieval from Scanned Forms Using Optical Character
Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Vaishali Aggarwal, Sourabh Jajoria and Apoorvi Sood
Combined Approach of Fuzzy Image Blocking Implementation for
Medical Image Fusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
Madhuri Patel and Drashti Baldev
A Systematic Review on Image Enhancement Techniques . . . . . . . . . . . . 227
Nishant Bhardwaj, Gagandeep Kaur and Pradeep Kumar Singh
Virtual Field and Wave Laboratory Using Scilab and Java . . . . . . . . . . 237
Sarika Agarwal and Gunjeet Kaur
Contents xiii

Classification and Comparative Study of IRS LISS-III Satellite Images

Using RBFN and Decision Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Anand Upadhyay, Santosh kumar Singh, Shailesh Kisan Gaikwad
and Ashutosh Chandra Mukherjee
An Analytic Review on Image Enhancement Techniques Based on Soft
Computing Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
Gagandeep Kaur, Nishant Bhardwaj and Pradeep Kumar Singh
An Auto-Threshold Control for Isolating False Shadow
from High-Resolution Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
K. Komathy
About the Editors

Dr. Shabana Urooj currently works at Gautam Buddha University, Greater Noida,
India. She received her bachelor’s in Electrical Engineering and M.Tech. in
Instrumentation and Control from Aligarh Muslim University, Aligarh, India. She
has completed her doctorate at Jamia Millia Islamia (A Central University), New
Delhi, India. She has 3 years of industrial and 15 years of teaching experience. She
has authored/co-authored more than 90 research papers. She has served as an editor,
board member, and reviewer for a number of leading journals. She is an active
volunteer of IEEE and actively involved in various professional development
ventures and bodies, such as the CSI and the ISTE.

Jitendra Virmani received his B.Tech. (Hons) in Instrumentation Engineering

from Sant Longowal Institute of Engineering and Technology, Punjab, in 1999; his
M.Tech. in Electrical Engineering with specialization in measurement and instru-
mentation from the Indian Institute of Technology Roorkee, India, in 2006; and his
Ph.D. on Analysis and Classification of B-Mode Liver Ultrasound Images from the
same institute in 2014. He held various academic posts before joining the
Department of Electrical and Instrumentation Engineering, Thapar University,
Patiala, Punjab, India, where he has been an Assistant Professor since July 2015. He
is a life member of the Institute of Engineers (IEI), India. His research interests
include application of machine learning and soft computing techniques for the
analysis of medical images.

xv
A Comparative Study of Various Color
Texture Features for Skin Cancer
Detection

Maurya Ritesh and Srivastava Ashwani

Abstract Detection of skin cancer gives the best chance of being diagnosed early.
Biopsy method for skin cancer detection is much painful. Human interpretation
contains difficulty and subjectivity; therefore, automated analysis of skin
cancer-affected images has become important. This paper proposes an automatic
medical image classification method to classify two major type skin cancers:
melanoma and non-melanoma. In this paper, we have used the color and texture
features in combination which gives better results than using color or gray-level
information alone. We have used k-means clustering algorithm to segment the
lesion. The features are extracted by seven different color texture feature extractors
from the segmented images. Classification accuracy of our proposed system is
evaluated on four different types of classifiers, and their values are compared with
one another. The results of the proposed system are computed on fivefolds of
cross-validation in order to perform better analysis of our proposed method.

Keywords Gray-level co-occurrence matrices Support vector machine Local




binary patterns Texture features Color percentiles K-means clustering





Co-occurrence matrix Color features Integrative co-occurrence matrix Gabor





features Linear classifier NN classifiers NMC classifiers Cross-validation

M. Ritesh (&)
Department of Computer Science and Engineering, Shri Ramswaroop
Memorial University, Lucknow, U.P, India
e-mail: [email protected]
S. Ashwani
Computer Engineering and Application Department, GLA University, Mathura, India
e-mail: [email protected]

© Springer Nature Singapore Pte Ltd. 2018 1

S. Urooj and J. Virmani (eds.), Sensors and Image Processing, Advances in Intelligent
Systems and Computing 651, https://doi.org/10.1007/978-981-10-6614-6_1
2 M. Ritesh and S. Ashwani

1 Introduction

Skin cancer incidence is increasing at 3.1% per year [1]. Skin cancer spread over
the body with the help of lymphatic and blood vessels. Thus, early detection of skin
cancer is very important for proper diagnosis of the disease.
Melanoma and non-melanoma are two major categories of skin cancers.
Malignant melanoma is of several subtypes. Basal cell carcinoma and squamous
cell carcinomas are two main types of non-melanoma skin cancers.
Each type of skin cancer is different from the other skin cancers in certain
characteristics.
In clinical detection of skin cancer diagnosis, dermatologists use a visual
inspection. Clinical diagnostic performance is very poor in comparison with der-
moscopy and automatic diagnosis. Dermoscopy is a noninvasive diagnostic tech-
nique. It uses clinical dermatology and dermatopathology in combination to inspect
the morphological features which is not possible in clinical detection. Dermoscopy
increases the performance of diagnosis with 10–30% compared to unaided eye [2].
Differentiation of skin cancer images needs much more experience with der-
moscopy technique. Less-experienced clinicians use ABCD-E rule [3] to improve
the diagnostic performance [3].
Automatic image processing of skin cancer gives better results by providing the
exact information about lesion, which can be useful for the clinician to detect and
classify skin cancer. It is also used as a stand-alone early warning tool. Effective
implementation of this automatic technique may give reduced deaths with benefits
both to the patients and to the healthcare system. Working of automatic skin cancer
detection has three main stages: (1) segmentation of lesion; (2) feature extraction
and feature selection; and (3) lesion classification.
Segmentation is an important process in image processing applications and
computer vision because doctors are always interested in meaningful regions of the
dermoscopy image. Segmentation divides an image into a number of separate
regions. Pixels, in each region, have high similarity such as color, intensity, and
texture. Many researchers use only gray level for image segmentation [31]. But, in
our proposed system, we use color information of the image for lesion
segmentation.
In general, we covert the color image in gray-level image; therefore, color
information does not used. There is a wide variety of segmentation methods used in
dermoscopy images [4]. Recent advancements include thresholding [5, 6], k-means
clustering [7], fuzzy c-means clustering [8, 9], density-based clustering [10],
mean-shift clustering [11], gradient vector flow snakes [12–14], color quantization
followed by spatial segmentation [15], statistical region merging [16], watershed
transformation [17], dynamic programming [18, 19], and supervised learning [20,
21]. Clustering is an unsupervised learning technique, where one can give the
number of clusters in advance to classify pixels [22]. A similarity measure is
defined between pixels, and similar pixels are then grouped into a cluster. We use
k-means clustering for segmentation of color images.
A Comparative Study of Various Color Texture Features for Skin … 3

It is very hard to differentiate skin cancer visually. Identification and extraction

of most effective features from cancer-affected lesion is very important. Each class
of skin cancer has some different features than others. We use these different
features for classification. Feature extraction extracts useful features or properties
from original dataset of an image. These extracted features easily classify the
classes of skin cancer.
Color features are mainly statistical parameters. These are calculated from inter-
and intra-channel of an image, like average value and standard deviation of the
RGB [23–27] or HSV color channels [28]. Here, we use “local binary pat-
terns + color percentiles,” “integrative co-occurrence matrices,” “gray-level
co-occurrence matrices + color percentiles,” “Gabor features + chromatic fea-
tures,” “Gabor features,” “opponent color LBP,” “color ranklets” [29]. These
methods are based on texture and disjoint color analysis. Textural features are
extracted from images by converting into gray level, and color features are com-
puted with the help of three color component of an image. Textural and color
features are concatenated into the same feature vector to improve the classification
accuracy.
The main aim of feature selection is to select the maximum number of features to
achieve high performance in cancer classification [30]. Feature selection is
important when anyone works on gray-level features. In our proposed system, there
is no need of feature selection algorithms.
Classification phase of the diagnostic system is the one in charge of making the
inferences about the extracted information in the previous phases in order to be able
to produce a diagnostic about the input image. In our experiment, we have used four
well-established classifiers: support vector classifier (SVC), nearest neighborhood
(NN), linear classifier, and nearest mean classifier (NMC).
The rest of the paper is organized as follows: Sect. 2 briefly reviews segmen-
tation, feature extraction techniques, and classification methods which are used in
proposed framework. Section 3 reports extensive experimental results, and Sect. 4
concludes the paper.

2 Proposed Framework

Proposed framework is a compiled abstraction of digital image classification.

Figure 1 describes the steps in an image classification.
In this paper, we propose this framework with k-means clustering segmentation
and color texture feature extraction techniques which is a new approach to classify
the skin cancer images. For validation purpose, twofold cross-validation is used.
4 M. Ritesh and S. Ashwani

Fig. 1 Proposed method

2.1 Segmentation

Segmentation is a process to partitioning an image into disjoint regions that are

homogeneous with respect to a chosen property such as luminance, color, and
texture. The aim of segmentation is to change the representation of an image into
something that is more meaningful and easier to analyze [30].
K-means Clustering Segmentation: K-means clustering method is one of the
simplest unsupervised learning methods and this method is nondeterministic,
numerical, and iterative. In this experiment, color images are used as input; hence,
A Comparative Study of Various Color Texture Features for Skin … 5

Fig. 2 Input image and

segmented image

this technique is used for segmentation. K-means clustering is partitioning method.

This method groups objects in the way that within-group variance is minimized. If
within-group variance is minimized, then it gives high featured segmented image.
The working of this method is as follows [31, 32]:
1. Select K pixels for centers randomly. Initially, center represents group centroids.
2. For grouping the sample data, calculate histogram bin value distance between
pixels and selected centroids and assign the group on the basis of nearest
distance.
3. Calculate the histogram bin value for new group to find the new position of
centroids.
4. If the value of centroids changes, then repeat steps 2 and 3 (Fig. 2).

2.2 Feature Extraction

It is important to identify the most effective features to extract from skin cancer
lesions and to find the most effective pattern-classification criteria and algorithms
for differentiating those lesions.
Local Binary Patterns + Color Percentiles: The LBP feature vector is created in
the following way:
1. Divide the window size into cells in the form of n*n matrix (i.e., 16*16).
2. If the center pixel value (Threshold Value) is less than the selected pixel, then
write 1, otherwise write 0. For rotational invariant features, move in one
direction either clockwise or anticlockwise. This step gives 8-digit binary
number. Convert it into decimal number for further processing.
3. Compute the histogram of the frequency of each “number” occurring over the
given cell.
4. Apply normalization on histograms.
5. Concatenate the histograms of all divided cells.
The combination of LBP + CP presented by Niskanen et al. [33]. In this method,
we calculate local binary patterns in an grayscale image. For better results, we
reduce the calculated features into rotationally invariant features. Rotational
invariance is necessary because when image is rotated, the gray values also rotated
in a circular form where origin is same. Then, calculation of feature vector changes
6 M. Ritesh and S. Ashwani

with the rotation. We apply discrete Fourier transformation function (DFT) to make
features rotationally invariant. There are always 36 rotationally invariant features in
a grayscale image. Now, we calculate the color percentiles of first, second, and third
quartiles (three points that divide the each channel into four parts) of three channels.
Total features from color percentiles are nine. When we concatenate features of
these two methods, the final resulting feature vector dimension is 36.
Integrative Co-occurrence Matrices: Integrative co-occurrence matrix [34–36] is
a new approach for calculating the inter-channel and intra-channel features. In
inter-channel feature calculation, we calculate the co-occurrence matrix features
from each of three channels, and in intra-channel, we calculate the features from
combination of color channels ((r,g), (r,b), (g,b)). In our experiment, we use five
co-occurrence features: energy, contrast, correlation, homogeneity, and inertia.
Final resulting feature vector dimension is 30.
Co-occurrence Matrices + Color Percentiles: The combination of co-occurrence
matrices and color percentiles is presented by Niskanen et al. [33]. This method is first
used for applications in wood inspection. Out of 14 features, we computed five
important co-occurrence features for classification. Color percentile is calculated as
discussed in LBPnCP method. So, the dimension of final resulting feature vector is 14.
Gabor Features: Gabor features are used to extract scale and orientation infor-
mation from different channels. We calculate Gabor features for inter-channel of an
image with above-mentioned configuration. One channel gives 36 rotationally
variant features. So, the resulting feature vector is of 32 + 32 + 32 = 96 dimension
when we apply DFT normalization to achieve rotationally invariant features.
Gabor Features + Chromatic Features: Different frequencies and orientations are
used in Gabor filters for extracting features from an image. The tuning of Gabor
filters with these parameters: number of orientations (no) = 6, number of frequen-
cies (nF) = 4, frequency ratio (Fr) = half-octave, standard deviation of Gaussian
envelope (longitudinal) ( ) = 0.5, standard deviation of Gaussian envelop
(transversal) ( ) = 0.5, max frequency (Fm) = 0.327. This tuning is same in overall
experiment.
Opponent Color LBP: The processing of opponent color LBP feature extraction
method is same as LBP method. In opponent color LBP method, we calculate features
from inter-channel and intra-channel, but in LBP method, we calculate features from
grayscale image. Separate channels give 108 (36 + 36 + 36) features, and paired
channels give 108 features. The dimension of resulting feature vector is 216.
Color Ranklets: Ranklet transform is an image processing technique which
considers the relative rank of neighboring pixels instead of their values. Invariance
against rotation is achieved by computing the discrete Fourier transform (DFT) of a
set of ranklet features computed from circular neighborhoods at different rotation
angles.
A Comparative Study of Various Color Texture Features for Skin … 7

2.3 Classifier

Nearest Neighbor and Nearest Mean Classifier: An input pattern is classified in the
class of the nearest training pattern (NN) or that of the nearest centroid (NMC) [37,
38]. Absence of parameter tuning makes these classifiers advantageous and easy to
implement. In NN, sensitivity to outliers makes it poor performer. In nearest mean
classifier, misclassifications arise because probably centroids are not representative.
Linear Classifier: Linear classifiers classify features by making a classification
decision which is based on linear combination of the feature values. Linear classifier is
originally developed for binary classification. It requires a predefined linear function
(hyperplane) that best separates the required classes in the feature space. If the two
classes are linearly separable, then perfect separation between classes exists.
Support Vector Classifier: support vector machine is highly effective classifier
and currently has a great importance in pattern recognition and artificial intelli-
gence. The tuning of a SVC is very important and needs very careful analysis. In
our experiment, we are using RBF kernel function in SVM classifier; then, we have
to tune two parameters: C and gamma (the radius of RBF).
C is used during the training phase and says how much outliers are taken into
account in calculating support vector. C is a trade-off between training error and the
flatness of the solution. If the value of C is larger, then the final training error will be
less. But if we increase value of C too much, then the risk of losing the general-
ization properties of the classifier is high. Large C increases the time needed for
training, and small C makes classifier flat. We have to find a C that keeps the
training error small and generalizes well. In SVC processing, we choose a kernel
function which mapped patterns into a high-dimensional space [39]. According to
Hsu et al. [40], if the number of features is not large, then RBF kernel is a rea-
sonable choice.

3 Experiments and Simulation

3.1 Database Preparation

In our experiment, we use database of skin cancer images (melanoma and

non-melanoma) have been used from University of Waterloo. The University has
collected these images from Internet source [41, 42]. Collection of 150 images is
used in this experiment. Seventy-five images are of melanoma and 75 of
non-melanoma type [43, 44].
Size of image is always an important aspect of image processing experiments. In
this experiment, input image is resized to 128*128, which is quite suitable reso-
lution for obtaining better results.
Resized images are segmentated using k-means clustering, which are trained and
tested on different classifiers.
8 M. Ritesh and S. Ashwani

3.2 Result Analysis

The size of feature vector extracted using these feature extraction algorithms is
shown below in Table 1. It is clear from the table increasing the dimensionality of
feature vector increases the computation time (Fig. 3).
Tables 2, 3, 4, 5, 6, 7, and 8 display classification accuracy of different classifiers
for the given feature set using twofold cross-validation approach by varying the
training and testing ratio.

Table 1 Elapsed time of methods in increasing order

Method Number of features Elapsed time (s) Reference
LBPnCP 45 54.533748 (M. Niskanen et al. 2001)
GLCM + CP 14 90.964714 (M. Niskanen et al. 2001)
Gabor + CF 36 110.209004 (A. Drimbarean et al. 2001)
Gabor 96 208.859732 (G. Paschos, 2001)
Integrated CM 30 393.141803 (V. Arvis et al. 2004)
Opponent color LBP 216 13781.195905 (T. Maenpaa et al. 2004)
Color ranklets 216 28463.284710 (F. Bianconi et al. 2009)

Fig. 3 Images from dataset first row: melanoma, second row: non-melanoma

Table 2 Accuracy of Training/testing NN NMC Linear SVC

LBPnCP feature extraction
technique on different (60,15) 65.56 50.00 63.33 62.00
classifiers with given training (55,20) 57.14 61.43 51.43 57.14
and testing ratio (50,25) 68.00 58.00 62.00 66.00
(45,30) 80.00 66.67 80.00 83.33
(40,35) 80.00 70.00 93.36 89.14
A Comparative Study of Various Color Texture Features for Skin … 9

Table 3 Accuracy of Training/testing NN NMC Linear SVC

Gabor + CF feature
extraction technique on (60,15) 62.00 66.67 63.33 68.89
different classifiers with given (55,20) 70.00 77.14 70.00 68.57
training and testing ratio (50,25) 76.00 80.00 70.00 82.00
(45,30) 86.67 46.67 86.67 80.00
(40,35) 90.00 80.00 80.00 90.00

Table 4 Accuracy of Training/testing NN NMC Linear SVC

integrated CM feature
extraction technique on (60,15) 70.00 66.67 53.33 66.67
different classifiers with given (55,20) 64.29 62.86 67.14 70.00
training and testing ratio (50,25) 84.00 80.00 74.00 72.00
(45,30) 86.67 83.33 83.33 86.67
(40,35) 90.00 70.00 80.00 87.76

Table 5 Accuracy of Training/testing NN NMC Linear SVC

GLCM + CP feature
extraction technique on (60,15) 60.00 57.78 58.89 64.44
different classifiers with given (55,20) 61.43 47.14 68.57 64.29
training and testing ratio (50,25) 74.00 84.00 82.00 82.00
(45,30) 80.00 76.67 83.33 83.33
(40,35) 91.24 60.00 88.65 93.94

Table 6 Accuracy of Gabor Training/testing NN NMC Linear SVC

feature extraction technique
on different classifiers with (60,15) 71.11 74.44 74.44 76.67
given training and testing (55,20) 64.29 71.43 54.29 67.14
ratio (50,25) 78.00 74.00 78.00 78.00
(45,30) 90.00 83.33 76.67 90.00
(40,35) 90.00 80.00 60.00 96.57

Table 7 Accuracy of Training/testing NN NMC Linear SVC

opponent color LBP feature
extraction technique on (60,15) 55.56 65.56 60.00 63.33
different classifiers with given (55,20) 67.14 68.57 64.29 68.57
training and testing ratio (50,25) 76.00 66.00 68.00 80.00
(45,30) 86.67 76.67 90.00 86.67
(40,35) 92.29 50.00 80.00 95.11
10 M. Ritesh and S. Ashwani

Table 8 Accuracy of color Training/testing NN NMC Linear SVC

ranklets feature extraction
technique on different (60,15) 68.89 57.78 60.00 63.33
classifiers with given training (55,20) 72.86 55.71 61.43 70.00
and testing ratio (50,25) 74.00 56.00 72.00 74.00
(45,30) 76.67 60.00 70.00 76.67
(40,35) 80.00 70.00 78.00 80.00

Average accuracies obtained from each classifier for different features are plotted
graphically in Fig. 4.
It is clear from Fig. 4 that Gabor feature performs better than the other features.
LBPnCP, integrated GLCM, CLCM + CP are the better choice, while the color
ranklets and opponent color LBP perform poorly in time and space domains.
LBPnCP and Gabor features are best suited for optimum results considering the
running time and accuracy of prediction.
Pairwise comparison between the algorithms is done for better visual compar-
ison (Fig. 5). The directions of the arrow denote the method which has more
prediction accuracy than the other.

Fig. 4 Average accuracy of different feature extraction algorithms

A Comparative Study of Various Color Texture Features for Skin … 11

Fig. 5 Pairwise comparison of the feature extraction algorithm efficiency

4 Conclusion

Experimental results show that color and texture descriptors for skin cancer clas-
sification provide good classification accuracy. We have also evaluated the per-
formance of four different classifiers on these extracted features. SVC(support
vector classifier) outperforms among all others with the same features set and then
comes linear classifier, whose classification accuracy is nearly same as the SVC.
1-NN performs poorly on the given feature sets.
Gabor features proved to be the best features that can be used for this particular
application. Other more suitable methods are LBPnCP and integrated GLCM which
are also giving the promising results.
Opponent and color ranklets are the methods which are not advisable to be used
for the particular application because of their large dimensions and less prediction
accuracy as well.
12 M. Ritesh and S. Ashwani

References

1. Linos, E., Swetter, S.M., Cockburn, M.G., Colditz, G.A., Clarke, C.A.: Increasing burden of
melanoma in the United States. J. Invest. Dermatol. 129(7), 1666–1674 (2009)
2. Mayer, J.: Systematic review of the diagnostic accuracy of dermatoscopy in detecting
malignant melanoma. Med. J. Aust. 167(4), 206–210 (1997)
3. Robinson, J.K., Turrisi, R.: Skills training to learn discrimination of ABCDE criteria by those
at risk of developing melanoma. Arch. Dermatol. 142(4), 447–452 (2006)
4. Celebi, M.E., Iyatomi, H., Schaefer, G., Stoecker, W.V.: Lesion border detection in
dermoscopy images. Comput. Med. Imaging Graph. 33(2), 148–153 (2009)
5. Iyatomi, H., Oka, H., Celebi, M.E., Hashimoto, M., Hagiwara, M., Tanaka, M., Ogawa, K.:
An improved internet-based melanoma screening system with dermatologist-like tumor area
extraction algorithm. Comput. Med. Imaging. Graph. 32(7), 566–579 (2008)
6. Garnavi, R., Aldeen, M., Celebi, M.E., Varigos, G., Finch, S.: Border detection in
dermoscopy images using hybrid thresholding on optimized color channels. Comput. Med.
Imaging Graph. 35(2), 105–115 (2011)
7. H. Zhou, M. Chen, L. Zou, R. Gass, L. Ferris, L. Drogowski, and J. M. Rehg, “Spatially
Constrained Segmentation of Dermoscopy Images,” Proce. 5th IEEE International
Symp. Biomed. Imaging, pp. 800–803, (2008)
8. Schmid, P.: Segmentation of digitized dermatoscopic images by two-dimensional color
clustering. IEEE Trans. Med. Imaging 18(2), 164–171 (1999)
9. Zhou, H., Schaefer, G., Sadka, A., Celebi, M.E.: Anisotropic mean shift based fuzzy C-means
segmentation of dermoscopy images. IEEE J. Sel. Top. Sign. Proces. 3(1), 26–34 (2009)
10. Mete, M., Kockara, S., Aydin, K.: Fast density-based lesion detection in dermoscopy images.
Comput. Med. Imaging Graph. 35(2), 128–136 (2011)
11. R. Melli, C. Grana, and R. Cucchiara, “Comparison of color clustering algorithms for
segmentation of dermatological images,” Proc. SPIE Med. Imaging Conf. pp. 1211–1219
(2006)
12. Erkol, B., Moss, R.H., Stanley, R.J., Stoecker, W.V., Hvatum, E.: Automatic lesion boundary
detection in dermoscopy images using gradient vector flow snakes. Skin Res. Technol. 11(1),
17–26 (2005)
13. Zhou, H., Schaefer, G., Celebi, M.E., Lin, F., Liu, T.: Gradient vector flow with nean shift for
skin lesion segmentation. Comput. Med. Imaging Graph. 35(2), 121–127 (2011)
14. Abbas, Q., Celebi, M.E., Garcia, I.F.: A novel perceptually-oriented approach for skin tumor
segmentation. Int. J. Innovative Comput. Inf. Control 8(3), 1837–1848 (2012)
15. Celebi, M.E., Aslandogan, Y.A., Stoecker, W.V., Iyatomi, H., Oka, H., Chen, X.:
Unsupervised border detection in dermoscopy images. Skin Res. Technol. 13(4), 454–462
(2007)
16. Celebi, M.E., Kingravi, H.A., Iyatomi, H., Aslandogan, A., Stoecker, W.V., Moss, R.H.,
Malters, J.M., Grichnik, J.M., Marghoob, A.A., Rabinovitz, H.S., Menzies, S.W.: Border
detection in dermoscopy images using statistical region merging. Skin Res. Technol. 14(3),
347–353 (2008)
17. Wang, H., Moss, R.H., Chen, X., Stanley, R.J., Stoecker, W.V., Celebi, M.E., Malters, J.M.,
Grichnik, J.M., Marghoob, A.A., Rabinovitz, H.S., Menzies, S.W., Szalapski, T.M.: Modified
watershed technique and Post-Processing for segmentation of skin lesions in dermoscopy
images. Comput. Med. Imaging Graph. 35(2), 116–120 (2011)
18. Abbas, Q., Celebi, M.E., Garcia, I.F., Rashid, M.: Lesion border detection in dermoscopy
images using dynamic programming. Skin Res. and Technol. 17(1), 91–100 (2011)
19. Abbas, Q., Celebi, M.E., Garcia, I.F.: Skin tumor area extraction using an improved dynamic
programming approach. Skin Res. Technol. 18(2), 133–142 (2012)
A Comparative Study of Various Color Texture Features for Skin … 13

20. Schaefer, G., Rajab, M.I., Celebi, M.E., Iyatomi, H.: Colour and contrast enhancement for
improved skin lesion segmentation. Comput. Med. Imaging Graph. 35(2), 99–104 (2011)
21. Wighton, P., Lee, T.K., Lui, H., McLean, D.I., Atkins, M.S.: Generalizing common tasks in
automated skin lesion diagnosis. IEEE Trans. Information Technol. Biomed. 15(4), 622–629
(2011)
22. V. K. Dehariya, S. K. Shrivastava, R. C. Jain, “Clustering of image data set using K-Means
and fuzzy K-Means Algorithms”, International conference on CICN, pp. 386–391, (2010)
23. J. F. Aitken, J. Pfitzner, D. Battistutta, O.’Rourke PK, A. C. Green, and N.G. Martin,
“Reliability of computer image analysis of pigmented skin lesions of Australian adolescents,”
J. Cancer, 78(2), 252–257 (1996)
24. Green, A., Martin, N., Pfitzner, J., O’Rourke, M., Knight, N.: Computer image analysis in the
diagnosis of melanoma. J. Am. Acad. Dermatol. 31(6), 958–964 (1994)
25. H. C. Lee, Skin cancer diagnosis using hierarchical neural networks and fuzzy logic,
department of computer science, University of Missouri, Rolla (1994)
26. Seidenari, S., Burroni, M., Dell’Eva, G., Pepe, P., Belletti, B.: Computerized evaluation of
pigmented skin lesion images recorded by a videomicroscope: comparison between polarizing
mode observation and oil/slide mode observation. Skin Res. Technol. 1(4), 187–191 (1995)
27. K. Ramlakhan and Y. Shang,: “A mobile automated skin lesion classification system,”
Proceedings of the 23rd IEEE International Conference on Tools with Artificial Intelligence
(ICTAI ’11), pp. 138–141, Nov (2011)
28. Cascinelli, N., Ferrario, M., Bufalino, R., et al.: Results obtained by using a computerized
image analysis system designed as an aid to diagnosis of cutaneous melanoma. Melanoma
Res. 2(3), 163–170 (1992)
29. F. Bianconia, E. Gonzalezb, A. Fernandezb, S.A. Saet-taa: “Automatic classification of
granite tiles through colour and texture features”, Preprint submitted to Expert Systems with
Applications, Feb 20, (2012)
30. A. Masood and A. Ali Al-Jumaily, “Computer Aided Diagnostic Support System for Skin
Cancer: A Review of Techniques and Algorithms” Hindawi Publishing Corporation
International Journal of Biomedical Imaging, Volume (2013)
31. V. Jumb, M. Sohani, A. Shrivas: “Color image segmentation using K-Means clustering and
Otsu’s adaptive thresholding,” Int. J. Innovative Technol. Exploring Eng. (IJITEE), February
(2014)
32. Ojala, T., Pietikainen, M., Maenpaa, T.: Multiresolution gray-scale and rotation invariant
texture classification with Local Binary Patterns. IEEE Trans. Pattern Anal. Mach. Intell. 24,
971–987 (2002)
33. Niskanen, M., Silven, O., Kauppinen, H.,: “Color and texture based wood inspection with
non-supervised clustering”, Proceedings of the 12th Scandivanian Conference on Image
Analysis. Bergen, pp. 336–342, (2001)
34. Arvis, V., Debain, C., Berducat, M., Benassi, A.: Generalization of the cooccurrence matrix
for colour images: application to colour texture classification. Image Anal. Stereology 23(3),
63–72 (2004)
35. Palm, C.: Colour texture classification by integrative co-occurrence matrices. Pattern Recogn.
37(5), 965–976 (2004)
36. Bianconi, F., Fernandez, A.: Evaluation of the effects of Gabor filter parameters on texture
classification. Pattern Recogn. 40, 3325–3335 (2007)
37. F. Bianconi, A. Fernandez, E. Gonzalez, and J. Armesto, Robust color texture features based
on ranklets and discrete Fourier transform. J. Electron. Imaging 18, 043012-1-8 (2009)
38. Masotti, M., Campanini, R.: Texture classification using invariant ranklet features. Pattern
Recogn. Lett. 29, 1980–1986 (2008)
39. Scholkopf, B., Smola, A.: “Learning with Kernels: Support Vector Machines, Regularization,
Optimization, and Beyond”, MIT Press, (2002)
40. Hsu, C., Chang, C., Lin, C.: A practical guide to support vector classification. Department of
Computer Science National Taiwan University, Tech. rep. (2010)
14 M. Ritesh and S. Ashwani

41. http://www.dermis.net
42. http://www.dermquest.com
43. B.Gohila vani, A.Selvaraj, M.Vijaya priya, N.Anandh: “Segmentation and Classification of
Skin Lesions Based on Texture Features”, Int. J. Eng. Res. Appl. Dec. (2014)
44. Maurya R., Singh S.K., Maurya A.K., Kumar A.: “GLCM and Multi Class Support vector
machine based automated skin cancer classification IEEE Conference”, Page(s):444—
447Print ISBN:978-93-80544-10-6 INSPEC Accession Number:14382783, pp. 5–7 Mar.
(2014)
Pattern Classification and Retrieval
of Content-Based Images—a Step
Towards Amelioration

Hemjot and Amitabh Sharma

Abstract The progressive web and computerized advancements have forced

endless increment in the measure of visual data accessible to users. This trend
prompted the advancement of exploration area where retrieval of images is done
through the content of information which became familiar as CBIR (content-based
image retrieval). CBIR frameworks are to a great extent utilized as a part of
medicinal picture annotation, face recognition systems, security frameworks and so
on. In this paper, we will discuss about an efficient system for retrieving images
faster since speed and precision are important as well as techniques to obtain better
classification of images. To conquer the issue of extensive number of features
extracted which obliges vast measure of memory and processing force, we need to
build a blend of 3 techniques (SURF, SVM and LDA) which best portray the
information with adequate precision. Hence, we are using dimensionality reduction
algorithm LDA in combination with SVM for the classification purpose and SURF
which is quick and robust interest point detector.

Keywords Content-based image retrieval (CBIR)

Pattern recognition




Classification Support vector machine (SVM) Colour histogram Speeded up



robust features (SURF) Linear discriminant analysis (LDA)

1 Introduction

A significant expansion has been done in the digital field for more than many years.
This particular development has motivated research in image databases, which were
virtually dismissed simply by conventional systems due to the enormous level of

Hemjot (&)
A. Sharma

Department of IT, Chandigarh Engineering College, Landran, Mohali
Punjab, India
e-mail: [email protected]
A. Sharma
e-mail: [email protected]

© Springer Nature Singapore Pte Ltd. 2018 15

S. Urooj and J. Virmani (eds.), Sensors and Image Processing, Advances in Intelligent
Systems and Computing 651, https://doi.org/10.1007/978-981-10-6614-6_2
Exploring the Variety of Random
Documents with Different Content
 "But after that——"

"He took the revolver out of the drawer and ran out through the French
window."

"But the shutters were found to be bolted on the inside," I argued, "when
they were examined by the police inspector."

"So they were," he admitted. "Miss Monica had already been in there
with young Gerald. They had seen to the shutters."

"Then you think that Monica knew?"

"Of course she did."

"Then her desire to prove Morley Thrall guilty——"

"Was partly hatred of him, and partly the desire to shield her brother,"
the funny creature concluded as he collected traps, his bit of string and his
huge umbrella. "Think it over; you will see that I am right. I am sorry for
those two, aren't you? But they are selling Brudenell Court, I understand,
and their mother's fortune has become theirs absolutely. They will go
abroad together, make a home for themselves, and one day, perhaps,
everything will be forgotten, and a new era of happiness will arise for the
innocent, now that the guilty has been so signally punished. But it was an
interesting case. Don't you agree with me?"

IX

THE MYSTERY OF THE WHITE CARNATION

 §1

"I suppose that is a form of snobbishness," the Old Man in the Corner
began abruptly.

I gave such a jump that I nearly upset the contents of a cup of boiling tea
which I was conveying to my mouth. As it was, I scalded my tongue and
nearly choked.

"What is?" I queried with a frown, for I was really vexed with the
creature. I had no idea he was there at all. But he only smiled and concluded
his speech, quite unperturbed.

"... that creates additional interest in a crime when it concerns people of

wealth or rank."

"Snobbishness," I rejoined, "of course it's snobbishness! And when the

little suburban madam has finished reading about Lady Stickinthemud's
reception at Claridge's she likes to turn to Lord Tomnoodle's prospective
sojourn in gaol."

"You were thinking of the disappearance of the Australian millionaire?"

he asked blandly.

"I don't know that I was," I retorted.

"But of course you were. How could any journalist worthy of the name
fail to be interested in that intricate case?"

"I suppose you have your theory—as usual?"

"It is not a theory," the creature replied, with that fatuous smile of his
which always irritated me; "it is a certainty."

Then, as he became silent, absorbed in the contemplation of a

wonderfully complicated knot in his beloved bit of string, I said with
gracious condescension:
 "You may talk about it, if you like."

He did like, fortunately for me, because, frankly, I could not see daylight
in that maze of intrigue, adventure and possibly crime, which was described
by the Press as "The Mystery of the White Carnation."

"The events were interesting from the outset," he began after a while,
whilst I settled down to listen, "and so were various actors in the society
drama. Chief amongst these was, of course, Captain Shillington, an
Australian ex-officer, commonly reputed to be a millionaire, who, with his
mother and sister, rented Mexfield House in Somerset Street, Mayfair, the
summer before last. It appears that Lord Mexfield's younger son, the
Honorable Henry Buckley, who was an incorrigible rake and whom his
father had sent on a tour round the world in order to keep him temporarily
out of mischief, not to say out of gaol, had met a married brother of Captain
Shillington's out in the Antipodes, they had been very kind to him, and so
on, with the result that when came the following London season the family
turned up in England, and, after spending a couple of days at the Savoy,
they moved into the Mexfields' house in Somerset Street.

"Lord and Lady Mexfield were abroad that year, and Henry Buckley and
his sister Angela were living with an aunt who had a small house
somewhere in Mayfair.

"Although the Shillingtons were reputed to be very wealthy, they

appeared to be very quiet, simple folk, and it certainly seemed rather
strange that they should have gone to the expense of a house in town, when
obviously they had no social ambitions and did not mean to entertain. As a
matter of fact, as far as Mrs. Shillington and her daughter were concerned,
nobody could have lived a quieter, more retiring life than they did. Mrs.
Shillington was an invalid and hardly ever went outside her front door, and
the girl Marion seemed to be suffering from a perpetual cold in the head.
They seemed to be in a chronic state of servant trouble. Mrs. Shillington
was dreadfully irritable, and one set of servants after another were engaged
only to leave without notice after a few days. The one faithful servant who
remained was a snuffy old man who came to them about a month after they
moved into Mexfield House. He and a charwoman did all the work of
cooking and valeting and so on. Presumably the old man could not have got
a situation elsewhere as his appearance was very unprepossessing, and
therefore he was willing to put up with what the servants' registry offices
would term 'a very uncomfortable situation.'

"Captain Shillington, the hero of the tragic adventure, on the other hand,
went about quite a good deal. He was certainly voted to be rather strait-
laced, not to say priggish, but he was very good-looking and a fine dancer.
Henry Buckley introduced him to some of his smart friends and Lady
Angela constituted him her dancing partner. The partnership soon
developed into warmer friendship and presently it was given out that Lady
Angela Buckley, only daughter of the Earl and Countess of Mexfield, was
engaged to Captain Denver Shillington, the Australian millionaire. Lady
Angela confided to her friends that her fiancé was the owner of immense
estates in Western Australia, on a portion of which rich deposits of gold had
lately been discovered. He certainly had plenty of money to spend, and on
one occasion he actually paid Henry Buckley's gambling debts to the tune
of two or three hundred pounds.

"On the whole, society pronounced the match a suitable one. Lady
Angela Buckley was no longer in her first youth, whilst her brother, to
whom she was really devoted, would be all the better for a somewhat
puritanical, strait-laced and, above all, wealthy brother-in-law."

§2

"That, then, was the position," the Old Man in the Corner continued after
a while, "and the date of Lady Angela Buckley's marriage to Captain
Denver Shillington had been actually fixed when the public was startled one
afternoon towards the end of the summer by the sensational news in all the
evening papers: 'Mysterious disappearance of a millionaire.' This highly
coloured description applied, as it turned out, to Captain Shillington, the
fiancé of Lady Angela Buckley. It seems that during the course of that same
morning a young lady, apparently in deep distress and suffering from a
streaming cold in the head, had called at Scotland Yard. She gave her name
and address as Marion Shillington, of Mexfield House, Somerset Street,
Mayfair, and stated that she and her mother were in the greatest possible
anxiety owing to the disappearance of her brother, Captain Denver
Shillington. They had last seen him on the previous Friday evening at about
nine o'clock when he left home in order to pick up his fiancée, Lady Angela
Buckley, whom he was escorting that night to a reception in Grosvenor
Square. He was wearing full evening dress and a soft hat. Miss Shillington
couldn't say whether he had any money in his pockets. She thought that
probably he was carrying a gold cigarette case, which Lady Angela had
given him, but, as a matter of fact, he never wore any jewellery.

"No one in the house had heard him come in again that night, and his
bed had not been slept in. Questioned by the police, Miss Shillington
explained that neither she nor her mother felt any alarm at first because
there had been some talk of Captain Shillington going away with his
fiancée to stay with friends over the week-end, somewhere near
Newmarket. It was only this morning, Wednesday, that Mrs. Shillington
first began to worry when there was still no sign or letter from him. 'My
brother is a very good son,' Miss Shillington continued, explaining to the
police, 'and always very considerate to mother. It was so unlike him to leave
us without news all this while and not let us know when to expect him
home. So I rang up Lady Angela Buckley, who is his fiancée, to see if I
could get news through her, as I could see mother was beginning to get
anxious. Mr. Henry Buckley, Lady Angela's brother, answered the 'phone. I
asked after his sister and he told me that she was staying on in the country a
day or two longer. He himself had come back to town the previous night. I
then asked him, quite casually, if he knew whether Denver—that's my
brother—would be returning with Angela. And his answer to me was,
"Denver? Why, I haven't seen him since Friday. And I can tell you that he is
in for a row with Angela. She was furious with him that he never wrote
once to her while she was away." I was so upset that I hung up the receiver
and just sat there wondering what to do next. But Mr. Buckley rang up a
moment or two later and asked quite cheerily if there was anything wrong.
"Good old Square-toes!" he said, meaning my brother, whom he always
used to chaff by calling him "Square-toes," "don't tell me he has gone off on
the spree without letting you know. I say, that's too bad of him, though. But
I shouldn't be anxious if I were you. Boys, you know, Miss Shillington, will
be boys, and I like old Square-toes all the better for it."'
 "Miss Shillington," the Old Man in the Corner went on, "was as usual
suffering from a streaming cold, and between spluttering and crying, she
had reduced two or three handkerchiefs to wet balls. At best she was no
beauty, and with a red nose and streaming eyes she presented a most
pitiable spectacle. 'I made Mr. Buckley assure me once more,' she said, 'that
he had seen nothing of Denver since Friday. That night he and Lady Angela
and Denver were at a reception in Grosvenor Square. They all left about the
same time. Angela and Denver went, presumably, straight home; at any rate,
he, Mr. Buckley, saw nothing more of them after they got into their car. He
himself went to spend an hour or two at his club and came home about two
a.m. The next morning, after breakfast, he drove his sister out to Tatchford,
near Newmarket, where they spent the week-end with some friends. And
that was all Mr. Buckley could say to me,' Miss Shillington concluded,
vigorously blowing her nose: 'He came home last night from Tatchford, and
was expecting Lady Angela in a couple of days. Denver had not been at
Tatchford at all, and he had not once written to Angela all the while she was
away.'

"Of course the police inspector to whom Miss Shillington related all
these facts had a great many questions to put to her. For one thing he
wanted to know whether she had been in communication with Lady Angela
Buckley since this morning.

"'No,' the girl replied, 'I have not, and so far, I haven't said anything to
mother. As soon as I felt strong enough I put on my things and came along
here.'

"Then the inspector wanted to know if she knew of any friends or

acquaintances of her brother's with whom he might have gone off for a
week-end jaunt without saying anything about it, either at home or to his
fiancée. He put the questions as delicately as he could, but the sister flared
up with indignation. It seems that the Captain's conduct had always been
irreproachable. He was a model son, a model brother, and deeply in love
with Lady Angela. Miss Shillington also refused to believe that he could
have been enticed to a place of ill-fame and robbed by one of the usual
confidence tricksters.
 "'My brother is exceptionally shrewd,' she declared, 'and a splendid
business man. Though he is not yet thirty, he has built up an enormous
fortune out in Australia, and administers his estates himself to the
admiration of every one who knows him. He is not the sort of man who
could be fooled in that way.'

"But beyond all this, and beyond giving a detailed description of her
brother's appearance, the poor girl had very little to say, and the detective
who was put in charge of the case could only assure her that enquiries
would at once be instituted in every possible direction, and that the police
would keep her informed of everything that was being done. Obviously, the
person most likely to be able to throw some light upon the mystery was
Lady Angela Buckley, but as you know, the advent of this charming lady
upon the scene only helped to complicate matters. It appears that Henry
Buckley, delighted at what he jocosely called, 'Old "Square-toes" falling
from grace,' had rung up his sister in order to tell her the startling news over
the telephone. Lady Angela being a very modern young woman, her brother
thought that she might storm for a bit but in the end see the humorous side
of the situation. But not at all! Lady Angela took the affair entirely au
tragique. Over the telephone she only exclaimed, 'Great Lord!' but at one
o'clock in the afternoon she arrived at the flat, having taken the first train up
to town and not even waiting for her maid to pack her things. Mr. Henry
Buckley was just going out to lunch. Without condescending to explain
anything, his sister dragged him off then and there to Scotland Yard.
'Something has happened to Denver,' was all that she would say. 'Something
dreadful, I am sure.' In vain did her brother protest that she would only be
making a fool of herself by rushing to the police like this, that old Square-
toes had only gone on the spree, and that, anyway, she ought to consult with
the Shillingtons before doing anything silly; Lady Angela would not listen
to reason. 'You don't know! You don't know!' she kept on reiterating with
ever-increasing agitation. 'He has been murdered, I tell you. Murdered!'

"By the time that the pair arrived at Scotland Yard, Lady Angela was in a
state bordering on hysterics, and her brother appeared both sulky and
perplexed. They saw the same Inspector who had interviewed Miss
Shillington, and certainly his amazement was no whit less than that of Mr.
Henry Buckley when Lady Angela having mentioned the disappearance of
Captain Denver Shillington, said abruptly, 'Yes, he has disappeared, and
incidentally, he had my pearls in his pocket.' The Inspector made no
immediate comment; men of his calling are used to those kinds of surprises,
but Henry Buckley gave a gasp of horror.

"'Your pearls?' he exclaimed. 'What pearls? Not——?'

"'Yes,' Lady Angela rejoined, coolly. 'The Glenarm pearls. All of them!'

"'But——' Henry Buckley stammered, wide-eyed and white to the lips.

"His sister threw him what appeared to be a warning glance, then she
turned once more to the police inspector.

"'My brother is upset,' she said calmly, 'because he knows that the pearls
are of immense value. The late Lord Glenarm left them to me in his will. He
made a huge fortune by a successful speculation in sugar. He had no
daughters of his own, and late in life he married my mother's sister. He was
my godfather, and when he first bought the pearls and gave them to his wife
as a wedding present, he said that after her death and his they should belong
to me. They were valued for probate at twenty-five thousand pounds.'

"Henry Buckley was still speechless, and it was in answer to several

questions put to her by the Inspector that Lady Angela gave the full history,
as far as she knew it, of the disappearance of her pearls.

"'I was going to spend the week-end with some friends at Tatchford, near
Newmarket,' she said. 'My brother at first had decided not to come with me.
On the Friday evening I went with Captain Shillington to a ball at the
Duchess of Flint's in Grosvenor Square. I wore my pearls; on the way home
in the car, Captain Shillington appeared very anxious as to what I should do
about the pearls whilst I was away. He wanted me to take them to the bank
first thing in the morning before I left. But I knew I couldn't do this,
because my train was at nine-fifty from Liverpool Street. Captain
Shillington had once or twice before shown anxiety about the pearls and
urged me to keep them at the bank when I was not wearing them, but he had
never been so insistent as that night.'
 "Lady Angela appeared to hesitate for a moment or two. She glanced at
her brother with a curious expression, both of anxiety and contempt. It
seemed as if she were trying to make up her mind to say something that was
very difficult, to put in so many words. The Inspector sat silent and
impassive, waiting for her to continue her story, and at last she did make up
her mind to speak.

"'I had a safe in the flat,' she went on, glibly, 'where I keep my jewellery,
but Captain Shillington did not seem satisfied. He argued and argued, and at
last he persuaded me to let him have the pearls while I was away and he
would deposit them at his own bank until my return.'

"Presumably at this point the lady caught an expression on the face of

the Inspector which displeased her, for she added with becoming dignity, 'I
am engaged to be married to Captain Denver Shillington.'

"'My God!' Henry Buckley exclaimed at this point, and with a groan he
buried his face in his hands.

"Mind you," the Old Man in the Corner proceeded, after a moment's
pause, "the public had no information as to the exact words, and so on, that
passed between Lady Angela, her brother Henry, and the officials of
Scotland Yard. All that I am telling you, and what I am still about to tell
you, came out bit by bit in the papers. Sensation-lovers were immensely
interested in the case from the outset, because, although both public and
police are familiar enough with the tragi-comedy of the good-looking
young blackguard who gets confiding females to entrust him with their little
bits of jewellery, this was the first time that the confidence trick had been
played by a well-known man about town—reputed wealthy, since he had
gone to the length of paying a friend's gambling debts—on a society lady
who was not in her first youth and must presumably have had some
knowledge of the world she lived in.

"Lady Angela had concluded her statements by saying that during the
drive home in the car she took off her pearls and handed them to her fiancé,
who slipped them into his pocket just as they were, although when presently
the car drew up at her door she suggested running up to her room to get the
case for them. The Captain, however, declared this to be unnecessary. What
he said was, 'I will sleep with them under my pillow to-night, and to-
morrow morning first thing I will take them round to the bank for you.'
After this he said good-night. Lady Angela let herself into the house with
her latchkey, and Captain Shillington then dismissed the car, saying that he
would enjoy a bit of a walk as the rooms at Grosvenor Square had been so
desperately hot.

"And it was at this point," the Old Man in the Corner now said with
deliberate emphasis as he worked away at an exceptionally intricate knot in
his beloved bit of string, "it was at this point that certain facts leaked out
which lent to the whole case a sinister aspect.

"It appears that on the Saturday morning at break of day one of the boats
belonging to the Thames District Police found a grey Homburg hat floating
under one of the old steamship landing stages and, stuck to one of the
wooden piles close by, a man's silk scarf. There was no name inside the hat
or any other clue as to the owner's identity, but both the scarf, which had
once been white or light grey, and the hat were terribly soiled and torn, and
both were stained with blood. The police had tried on the quiet to trace the
owner of the hat and scarf but without success. After Lady Angela had told
her story of the missing pearls, the things were shown to Miss Shillington,
who at once identified the hat as belonging to her brother; the scarf,
however, she knew nothing about.

"But this was not by any means all. It appears that for some reason
which was never quite clear, Captain Shillington, after he said good-night to
Lady Angela, altered his mind about the proposed walk. It may have started
to rain, or he may not, after all, have liked the idea of walking about the
streets at night with twenty-five thousand pounds' worth of pearls in his
pocket. Be that as it may, he hailed a passing taxi and drove to Mexfield
House. The driver came forward voluntarily in answer to an advertisement
put in the papers by the police. He stated that he remembered the
circumstance quite well because of what followed. He remembered taking
up a fare outside Stanhope Gate and being ordered to drive to Mexfield
House in Somerset Street. When he slowed down close to Mexfield House
he noticed a man with his hands in his pockets lounging under the doorway
of one of the houses close by. As far as he could see the man was in evening
dress and wore a light overcoat. He had on a silk hat tilted right over his
eyes so that only the lower part of his face was visible, and he had a white
or pale grey scarf tied loosely round his neck. The chauffeur also noticed
that he had a large white flower, probably a carnation, in his buttonhole.
After the taxi-man had put down his fare he drove off, and as he did so he
saw the man in the light overcoat step out from under the doorway, where
he had been lounging, and turn in the direction of Mexfield House. What
happened after that he didn't know, as he drove away without taking further
notice, but the police were already in touch with another man who had been
watching that night in Somerset Street, where a portion of the road was up
for repair. This man, whose name, I think, was William Rugger,
remembered quite distinctly seeing a 'swell' in a light overcoat and wearing
a light-coloured scarf round his neck, loafing around Mexfield House. He
remembered the taxi drawing up and a gentleman getting out of it,
whereupon the one in the light overcoat and the scarf went up to him and
said, 'Hullo, Denver!' at which the other gent, the one who had come in the
taxi, appeared very surprised, for Rugger heard him say, 'Good Lord, Henry,
what are you doing here?'

"Rugger didn't hear any more because the gentleman in the light
overcoat then took the other one by the arm and together the pair of them
walked away down the street. When they had gone Rugger noticed a large
white carnation lying on the pavement; he picked it up and subsequently
took it home to his missis.

"You may imagine what a stir and excitement this story—which pretty
soon leaked out in all its details—caused amongst the public. It seems that
although neither the taxi-driver nor the man Rugger had seen the face of the
man who had stepped out from under a neighbouring doorway and accosted
Captain Shillington, they were both of them quite positive that he was in
evening dress, and that he wore a silk hat, a light overcoat, and had a pale
grey or white scarf wound round his neck. And besides that, there was the
white carnation. But, of course, the crux of the whole evidence was
Rugger's assertion that he heard one gentleman—the one who got out of the
cab—say to the other in tones of great surprise, 'Good Lord, Henry, what
are you doing here?' Questioned again and again he never wavered in this
statement. He heard the name Henry quite distinctly and it stuck in his mind
because his eldest boy was Henry. He was also asked whether the
gentleman, who had stepped out of the taxi—obviously Captain Shillington,
since the other had called to him, 'Hullo, Denver'—walked away reluctantly
or willingly when he was thus summarily taken hold of by the arm. Rugger
was under the impression that he walked away reluctantly; he freed his arm
once, but the other got hold of him again, and, though Rugger did not catch
the actual words, he certainly thought that the two gentlemen were
quarrelling.

"And thus public opinion, which at first had been dead against the
Australian Captain, now went equally dead against Henry Buckley. Ugly
stories were current of his extravagance, his gambling debts, his addiction
to drink. People who knew him remembered one or two ugly pages in his
life's history: altercations with the police, raids on gambling clubs of which
he was a prominent member; there was even a fraudulent bankruptcy which
had been the original cause of his being sent out to Australia by his
harassed parents until the worst of the clouds had rolled by.

"The only thing that told in his favour, as far as the public was
concerned, was the bitter vindictiveness displayed against him by Miss
Shillington. That the girl had cause for bitterness was not to be denied. For
a time, at any rate, public opinion had branded her brother as a common
trickster and a thief, and she and her mother had no doubt suffered terribly
under the stigma; in consequence of this, Mrs. Shillington's health, always
in a precarious state, had completely broken down and the old lady had
taken to her bed, not suffering from any particular disease, but just from
debility of mind and body, obstinately refusing to see a doctor, declaring
that nothing would cure her except the return of her son.

"And on the top of all that came the growing conviction that the son
never would return and that he had been foully murdered for the sake of
Lady Angela's pearls, which he so foolishly was carrying in his pocket that
night. No wonder, then, that his sister Marion felt bitter against the people
who were the original cause of all these disasters; no wonder that she threw
herself heart and soul into the search for evidence against the man whom
she sincerely believed to be guilty of a most hideous crime.
 "It was mainly due to her that the police came on the track of William
Rugger, the night-watchman, and through the latter that the driver of the
taxi-cab was advertised for, because Rugger remembered seeing the
gentleman alight from a taxi outside Mexfield House. But Miss
Shillington's valuable assistance in the matter of investigation went even
further than that. She at last prevailed upon the old man-servant at Mexfield
House to come forward like a man and to speak the truth. He was a poor
creature, not really old, probably not more than fifty, but timid and almost
abject. He had at first declined to make any statement whatever, declaring
that he had nothing to say. To every question put to him by the police, he
gave the one answer, 'I saw nothing, sir, I 'eard nothing. I went to bed as
usual on the Friday night. The Captain 'e never expected me to sit up for 'im
when 'e was out to parties, and I never 'ear 'im come in, as I sleep at the top
of the 'ouse. No, sir, I didn't 'ear nothing that night. The last I seed of the
Captain was at nine o'clock, when 'e got into the car and said good-night to
me.' When he was shown the blood-stained hat, he burst out crying, and
said, 'Yes, sir! Yes, sir! That is the Captain's 'at. My Lord! What 'as become
of 'im?' He also failed to identify the scarf as being his master's property.

"Then one day Miss Shillington, still suffering from a cold in the head,
but otherwise very business-like and brisk, arrived at Scotland Yard with
the man—James Rose was his name—in tow. By what means she had
persuaded him to speak the truth at last no one ever knew, but in a
tremulous voice and shaken with nervousness, he did tell what he swore to
be the truth. 'I must 'ave dropped to sleep in the dining-room,' he said. 'I was
very tired that evening, and I remember after I 'ad cleared supper away I
just felt as 'ow I couldn't stand on my legs any longer, and I sat down in an
armchair and must 'ave dozed off. What woke me was the front-door bell
which rings in the 'all as well as in the basement. I looked at the clock, it
was past midnight. Captain forgot 'is key, that's what I thought. Lucky I
'adn't gone to bed, or I should never 'ave 'eard 'im. Funny 'is forgetting 'is
key, I thought. Never done such a thing before, I thought, and went to open
the door for 'im. But it wasn't the Captain,' Rose went on, his voice getting
more and more husky as no doubt he realised the deadly importance of what
he was about to say. 'No, it wasn't the Captain,' he reiterated, and shook his
head in a doleful manner.
 "'Who was it?' the Inspector demanded.

"'The young gentleman who sometimes came to the 'ouse,' Rose repeated
under his breath. 'Mr. 'Enery Buckley it was, sir. Yes, Mr. 'Enery, that's 'oo it
was.'

"'What did he say?' Rose was asked.

"''E asked if the Captain was in, and I said no, not as I knew, but I would
go and see. So up I went to the Captain's room and saw 'e wasn't there. Not
yet. And I told Mr. 'Enery so when I came down again.'

"'Then what happened?'

"'Mr. 'Enery 'e told me that 'e wouldn't wait and that I was to tell the
Captain 'e 'ad called, and that 'e would call again in 'arf an hour. I said that I
was going to bed and I wouldn't probably see the Captain. 'E might be ever
so late. Then Mr. 'Enery 'e just said, "Very good," and "Never mind," and
"Good-night, Rose," 'e said, and then I let 'im out.'

"'Well? And what happened after that?'

"'I don't know, sir,' the old man concluded. 'I went to bed and I never
seed the Captain again, nor yet Mr. 'Enery—not from that day to this, sir.
No, not again, sir.' And Rose once more shook his head in the same doleful
manner. Of course the police were very down on him for keeping back this
valuable piece of information, and they were even inclined to look with
suspicion upon the man. They wanted to know something about his
antecedents and why he seemed so frightened of facing the police
authorities. Fortunately for him, however, Miss Shillington could give them
all the information they wanted. She said that James Rose had been for
years in the service of a Mrs. O'Shea, who was a great friend of Mrs.
Shillington's. When Mrs. O'Shea died she left him a hundred pounds. But
the poor thing had never been very strong, and he was nothing to look at, he
couldn't get another place, and the hundred pounds vanished bit by bit.
About a month ago Mrs. Shillington, who was requiring a man-servant,
advertised for one in the Daily Mail. Rose answered the advertisement, and
though the poor thing in the meanwhile had gone terribly downhill
physically, Mrs. Shillington, remembering how honest and respectable he
had always been when he was in Mrs. O'Shea's service, engaged him out of
compassion and for the sake of old times. Miss Shillington gave him an
excellent character and the police were satisfied.

"I think," the Old Man in the Corner said, amorously contemplating a
marvellously intricate knot, which he had just made in his bit of string, "I
think that the police were mainly satisfied because at last they felt that 'they
had made out a case.' From that moment the detectives and inspectors in
charge became absolutely convinced that Henry Buckley had enticed
Captain Denver Shillington to some place of evil fame close to the river and
there, in collusion probably with other disreputable characters, had robbed
and murdered him. To say the least, the case looked black enough against
Buckley. His fast living, his mountain of debt, the absence in him of moral
rectitude as proved by his fraudulent bankruptcy, all told against him; and
now it was definitely proved that he had sought out and actually been in the
company of Captain Shillington the night that the latter disappeared. A light
grey overcoat similar to the one described by Rugger and by the chauffeur
as worn by the gentleman who was loafing in Somerset Street was found to
be a part of his wardrobe; no one could swear, however, as to the scarf, but
it turned out that he never went out in the evening without wearing a large,
white carnation in his button-hole.

"The fact that he had not stated from the beginning that he had called at
Mexfield House that night, and subsequently met the missing man and
walked away with him, naturally told terribly against him. Obviously the
man lost his head. Questioned by the police, he tried at first to deny the
whole thing: he declared that the man with the white carnation and the
light-coloured scarf was some other man whose name happened to be
Henry, and he tried to upset Rose's evidence by declaring that the man lied
and that he had never called at Mexfield House that night. But,
unfortunately for him, he had taken a taxi from his club to the house, the
taxi-driver was found, and the noose was further tightened round the
Honourable Henry Buckley's neck. In vain did he assert after that that
Denver Shillington had told him to call at Mexfield House at a quarter-past
midnight on that fatal Friday. He was no longer believed. He admitted that
he was in financial difficulties, and that he had spoken about these to
Captain Shillington earlier in the evening. He admitted, tardily enough, that
he went to Mexfield House hoping that Denver would give him some
money in order to wipe out his most pressing debts. When he found that the
Captain had not yet come home, he left a message with the man-servant and
thought he would go on to the club for a little while and return later to see
Shillington. Unfortunately, he drank rather heavily whilst he was at the club
and never thought any more either about his money worries or about the
Captain. In fact, he remembered nothing very clearly beyond the fact that
he went home, in the small hours and went straight to bed.

"He then went on to say that he woke up the next morning with a
splitting headache. It was pouring with rain and London was looking
particularly beastly, as he picturesquely termed it. He recollected that his
sister Angela had planned to go down with old Square-toes to some friends
near Newmarket for the weekend. He, too, had been asked but had declined
the invitation, but now he began to wish he hadn't; while he was out of town
money-lenders couldn't dun him, and a breath of country air would certainly
do him good.

"And he was just cogitating over these matters at eight a.m. on that
Saturday morning, when his sister Angela came into his room. 'She told
me,' he went on, 'that old Square-toes was unable to accompany her to these
friends in Cambridgeshire, that she didn't want to go alone, and would I hire
a car and drive her down. She offered to pay for the car, and, as the scheme
happened to suit me, I agreed. We drove down to Tatchford, and on the
Tuesday I had an unpleasant reminder from one of my creditors and thought
that I must get back to see what old Square-toes would do for me. I got
home that same evening, and the next morning early Miss Shillington rang
up and told me over the 'phone that they had heard nothing of Captain
Shillington since the previous Friday and that they were getting anxious.
And that's all I know,' he concluded. 'I swear that I never set eyes on
Shillington after he drove off from the Duchess of Flint's, with my sister in
his car. I did call at Mexfield House, but it was at Shillington's suggestion,
but when the man told me that the Captain was not yet home, I did not loaf
about the street, I went straight back to the club and then home.'
 "Of course all this was very clear and very categorical, but there were
one or two doubtful points in Buckley's statements, which the police—dead
out now to prove him guilty of murder—made the most of. Firstly, there
was his former denial on oath that he had not called at Mexfield House that
night. It was only when he was confronted with the testimony of the taxi-
cab driver that he made the admission. The employees at his club, which,
by the way, was in Hanover Square, had seen him come in at about half-
past eleven. He went out again twenty minutes later and the hall porter saw
him hail a taxi-cab. He was once more in the club at half-past twelve, and it
is a significant fact that two of the younger members chaffed him
subsequently because he had not the usual white carnation in his button-
hole.

"Then again it was more than strange that on the Friday he was so
worried about his debts that he went in the middle of the night to his friend's
house in order to try and borrow money from him, and yet when, according
to his own statements, he never even saw his friend, off he went the very
next morning to the country, stayed away four days, and on his return did
not make any attempt seemingly to see the Captain or to ask him for money.
Thirdly, it was equally inconceivable that Captain Shillington should have
appointed to see Buckley at that hour of the night, however pressed the
latter might have been for money. Why should he? The next morning would
have done just as well, whether he meant to help him or whether he did not,
and, according to the testimony of the night-watchman, William Rugger,
when he was accosted by Buckley, he exclaimed in tones of great surprise,
'Good Lord, Henry, what are you doing here?' These are not words which a
man would say to a friend whom he had appointed to meet at this very hour.

"However, this portion of the taxi-driver's and Rugger's testimony

Buckley still strenuously denied. He could not deny the other. He had called
at Mexfield House and reluctantly admitted that it had been nothing but
'blue funk' that had prompted him at first to hold his tongue about that and
then to deny the fact altogether.

"But, above all, there was yet another fact which to the police was more
conclusive, more damning than any other and that was that on the
Wednesday morning the Honourable Henry Buckley had called at Messrs.
Foster and Turnbull, the well-known pawnbrokers of Oxford Street, and had
pledged a pair of diamond ear-rings and a couple of valuable bracelets there
for which he received three hundred and fifty pounds.

"Here again, if Buckley had volunteered this statement, all might have
been well, but it was the pawnbrokers who gave information to the police. It
turned out that the ear-rings and the two bracelets were the property of his
sister, Lady Angela. Buckley declared that she had given them to him, and
she, very nobly, did her best to corroborate this statement of his, but it had
become impossible to believe a word he said. Lady Angela's valiant efforts
on his behalf were thought to be unconvincing, and, as a matter of fact, the
public has never known from that day to this whether Henry Buckley stole
his sister's jewellery, or whether she gave it to him voluntarily.

§3

"Mind you, there can be no question but that the police acted very
injudiciously when they actually preferred a charge of murder against
Henry Buckley. There were two such damning flaws in the chain of
evidence that had been collected against him that the man ought never to
have been arrested. Even the magistrate was of that opinion. As you know,
if there is the slightest doubt about such a serious charge, the magistrates
will always commit a man for trial and let a jury of twelve men pronounce
on the final issue rather than decide such grave matters on their own. But in
this case there were really no proofs. There were deductions: the accused
was a young blackguard, a moral coward and a liar. There was the blood-
stained scarf, the hat and the white carnation, there was the testimony of the
taxi-driver and the night watchman that Henry Buckley had been in the
company of Captain Shillington that night, but there was no proof that he
had murdered his friend and stolen the pearls.

"To begin with, if there had been a murder, where was it committed, and
what became of Captain Shillington's body? Of course, the police still hope
to find traces of it, but, as you know, they have not yet succeeded. Various
theories are put forward that Henry Buckley was a member of a gang of
ruffians with headquarters in some obscure corner of London close to the
river, and that he enticed the Captain there and murdered him with the help
of his criminal associates with whom he probably shared the proceeds of
the crime. But over a year has gone by since Shillington disappeared and
the police are no nearer finding the body of the missing man.

"The magistrate dismissed the case against Henry Buckley. There was
not sufficient evidence to commit him for trial. What told most in his favour
in the end was the question of time. He was able to prove that he was at his
club in Hanover Square at half-past midnight on the fateful night. Now,
according to James Rose's testimony, it was after midnight when he,
Buckley, called at Mexfield House. Even supposing that Shillington had
arrived in the taxi five minutes later, it was inconceivable that a man could
entice another to an out-of-the-way part of London, murder him—even if he
left others to dispose of the body—and walk back unconcernedly to
Hanover Square, all in less than half an hour. Nor were the pearls or any
large sum of money ever traced to Henry Buckley. He was just as deeply in
debt after the disappearance of Captain Shillington as he had been before.
Now he has gone on another tour round the world, and the Shillingtons—
mother and daughter—have given up all hopes of ever seeing the gallant
Captain, who was such a model son, again. A little while ago the illustrated
papers published photos of the two ladies on board a P. and O. steamer
bound for Australia, but the public had forgotten all about Lady Angela's
pearls and the mysterious white carnation. No one was interested in the old
lady with the white hair and stooping figure, who was carried on board in a
chair, and who obstinately refused to be interviewed by newspaper men
eager for copy. The case is relegated, as far as the public is concerned, to
the category of undiscovered crimes."

"But," I argued, as the Old Man in the Corner became silent, absorbed in
the untying of an intricate knot which he had made a little while ago,
"surely the police have found out who the man was who accosted Captain
Shillington in Somerset Street that night, the man with the light-coloured
scarf, which was subsequently found in the river by the side of the missing
man's hat, the man who called the Captain 'Denver,' and whom the latter
called 'Henry,' and was so surprised to see. If it was not Henry Buckley,
who was it?"
 "Ah!" the exasperating creature retorted with a fatuous smile, "who was
it? That's just the point—a point just as dark as that a man like Captain
Shillington could be enticed at that hour of the night to an out-of-the-way
part of London, and at a moment when he had his fiancée's jewellery worth
twenty-five thousand pounds in his pocket. Don't you think that that point is
absolutely inconceivable?"

"Well," I said, "it does seem——"

"Of course it does," he broke in eagerly. "I ask you: Is it likely? At one
moment we are told that Captain Shillington was a pattern of all the virtues
and that his business acumen and abilities had earned for him not only a
fortune but the admiration of all those who knew him; and the very next we
are asked to suppose that he would meekly allow a young blackguard,
whom he knew to be dishonest and unscrupulous, to drag him 'reluctantly'
to some obscure haunt of a gang of criminals. Surely that should have
jumped to the eyes of any sane person who had studied the case."

"I don't suppose," I retorted, "that Captain Shillington allowed Buckley

to drag him very far. Most people believed at the time that he was attacked
directly he rounded the corner of Somerset Street. There are one or two
entrances to mews just about there——"

"Yes," the funny creature rejoined excitedly, "but not one nearer than
fifty yards from Mexfield House. And do you think that the immaculate
Australian would have walked ten at night with young Buckley and with
those pearls in his pocket? Why should he? He was outside his own door.
Wouldn't he have taken Henry into the house with him if he wished to speak
to him? No! No! The whole theory is inconceivable...."

"But Captain Shillington disappeared," I argued, "and so did the pearls,

and his hat was found floating in the river, torn and blood-stained. You
cannot deny that."

"I certainly cannot deny," he replied, "that a blood-stained hat will float
on the water if it is thrown—say, from a convenient bridge."

"But the scarf?" I retorted.

Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com