Computer Vision Graphics and Image

Processing ICVGIP 2016 Satellite

Workshops WCVA DAR and MedImage
Guwahati India December 19 2016
Revised Selected Papers 1st Edition
Snehasis Mukherjee Et Al. (Eds.)
Visit to download the full and correct content document:
https://textbookfull.com/product/computer-vision-graphics-and-image-processing-icvgi
p-2016-satellite-workshops-wcva-dar-and-medimage-guwahati-india-december-19-20
16-revised-selected-papers-1st-edition-snehasis-mukherjee-et-al-eds/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Computer Vision Applications Third Workshop WCVA 2018

Held in Conjunction with ICVGIP 2018 Hyderabad India
December 18 2018 Revised Selected Papers Chetan Arora

https://textbookfull.com/product/computer-vision-applications-
third-workshop-wcva-2018-held-in-conjunction-with-
icvgip-2018-hyderabad-india-december-18-2018-revised-selected-
papers-chetan-arora/

Internetworked World 15th Workshop on e Business WeB

2016 Dublin Ireland December 10 2016 Revised Selected
Papers 1st Edition Ming Fan Et Al. (Eds.)

https://textbookfull.com/product/internetworked-world-15th-
workshop-on-e-business-web-2016-dublin-ireland-
december-10-2016-revised-selected-papers-1st-edition-ming-fan-et-
al-eds/

Computer Vision Pattern Recognition Image Processing

and Graphics Renu Rameshan

https://textbookfull.com/product/computer-vision-pattern-
recognition-image-processing-and-graphics-renu-rameshan/

Knowledge Engineering and Knowledge Management EKAW

2016 Satellite Events EKM and Drift an LOD Bologna
Italy November 19 23 2016 Revised Selected Papers 1st
Edition Paolo Ciancarini
https://textbookfull.com/product/knowledge-engineering-and-
knowledge-management-ekaw-2016-satellite-events-ekm-and-drift-an-
lod-bologna-italy-november-19-23-2016-revised-selected-
Autonomous Agents and Multiagent Systems AAMAS 2016
Workshops Best Papers Singapore Singapore May 9 10 2016
Revised Selected Papers 1st Edition Nardine Osman

https://textbookfull.com/product/autonomous-agents-and-
multiagent-systems-aamas-2016-workshops-best-papers-singapore-
singapore-may-9-10-2016-revised-selected-papers-1st-edition-
nardine-osman/

Image and Graphics Technologies and Applications 14th

Conference on Image and Graphics Technologies and
Applications IGTA 2019 Beijing China April 19 20 2019
Revised Selected Papers Yongtian Wang
https://textbookfull.com/product/image-and-graphics-technologies-
and-applications-14th-conference-on-image-and-graphics-
technologies-and-applications-igta-2019-beijing-china-
april-19-20-2019-revised-selected-papers-yongtian-wang/

Introduction to Visual Computing: Core Concepts in

Computer Vision, Graphics, and Image Processing 1st
Edition Aditi Majumder

https://textbookfull.com/product/introduction-to-visual-
computing-core-concepts-in-computer-vision-graphics-and-image-
processing-1st-edition-aditi-majumder/

Business Information Systems Workshops: BIS 2016

International Workshops, Leipzig, Germany, July 6-8,
2016, Revised Papers 1st Edition Witold Abramowicz

https://textbookfull.com/product/business-information-systems-
workshops-bis-2016-international-workshops-leipzig-germany-
july-6-8-2016-revised-papers-1st-edition-witold-abramowicz/

Graph Drawing and Network Visualization 24th

International Symposium GD 2016 Athens Greece September
19 21 2016 Revised Selected Papers 1st Edition Yifan Hu

https://textbookfull.com/product/graph-drawing-and-network-
visualization-24th-international-symposium-gd-2016-athens-greece-
september-19-21-2016-revised-selected-papers-1st-edition-yifan-
 Snehasis Mukherjee · Suvadip Mukherjee
Dipti Prasad Mukherjee · Jayanthi Sivaswamy
Suyash Awate · Srirangaraj Setlur
Anoop M. Namboodiri · Santanu Chaudhury (Eds.)

Computer Vision,
LNCS 10481

Graphics, and
Image Processing
ICVGIP 2016 Satellite Workshops, WCVA, DAR, and MedImage
Guwahati, India, December 19, 2016
Revised Selected Papers

123
Lecture Notes in Computer Science 10481
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen

Editorial Board
David Hutchison
Lancaster University, Lancaster, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Friedemann Mattern
ETH Zurich, Zurich, Switzerland
John C. Mitchell
Stanford University, Stanford, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
C. Pandu Rangan
Indian Institute of Technology, Madras, India
Bernhard Steffen
TU Dortmund University, Dortmund, Germany
Demetri Terzopoulos
University of California, Los Angeles, CA, USA
Doug Tygar
University of California, Berkeley, CA, USA
Gerhard Weikum
Max Planck Institute for Informatics, Saarbrücken, Germany
More information about this series at http://www.springer.com/series/7412
Snehasis Mukherjee Suvadip Mukherjee
•

Dipti Prasad Mukherjee Jayanthi Sivaswamy

•

Suyash Awate Srirangaraj Setlur

•

Anoop M. Namboodiri Santanu Chaudhury (Eds.)

•

Computer Vision,
Graphics, and
Image Processing
ICVGIP 2016 Satellite Workshops, WCVA, DAR, and MedImage
Guwahati, India, December 19, 2016
Revised Selected Papers

123
Editors
Snehasis Mukherjee Suyash Awate
Indian Institute of Information Technology Indian Institute of Technology Bombay
SriCity Mumbai
India India
Suvadip Mukherjee Srirangaraj Setlur
GE Global Research CEDAR
Bangalore Buffalo, NY
India USA
Dipti Prasad Mukherjee Anoop M. Namboodiri
Indian Statistical Institute International Institute of Information
Kolkata Technology
India Hyderabad
India
Jayanthi Sivaswamy
International Institute of Information Santanu Chaudhury
Technology CSIR-CEERI
Hyderabad Pilani, Rajasthan
India India

ISSN 0302-9743 ISSN 1611-3349 (electronic)

Lecture Notes in Computer Science
ISBN 978-3-319-68123-8 ISBN 978-3-319-68124-5 (eBook)
https://doi.org/10.1007/978-3-319-68124-5

Library of Congress Control Number: 2017956086

LNCS Sublibrary: SL6 – Image Processing, Computer Vision, Pattern Recognition, and Graphics

© Springer International Publishing AG 2017

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 International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
 Preface

The tenth version of the Indian Conference on Vision, Graphics and Image Processing
(ICVGIP) was held at IIT Guwahati in December 2016. ICVGIP already taken a unique
position within the scientific community in India. Starting with the ninth version of
ICVGIP, workshops on contemporary topics have become an integral part of the
conference. The most recent in this series of ICVGIP workshops were held on
December 19, 2016, at the IIT Guwahati. Four tracks on Computer Vision Applica-
tions, Medical Image Processing, Document Analysis and Recognition, and Digital
Heritage were part of the workshop program this year. The collection of papers in this
edited volume is extended versions of papers presented in the first three of the four
above mentioned tracks. The authors of the papers presented in this workshop had
submitted an extended version of the paper keeping in mind the feedback received at
the workshop.
Overall 52 papers were submitted in the Computer Vision Applications track and
after a rigorous double-blind review process, we could accept 18 papers keeping the
quality and time-frame of a single-day workshop. This track mainly focused on
industrial applications of computer vision and related technologies.
The Medimage Workshop focused on problems in medical image computing and
invited original contributions to address these problems. The workshop received 14
papers and after a double-blind review process, 7 were accepted for presentation as
poster or oral.
The DAR Workshop was organized to bring together people working in the field of
Document Analysis and Recognition. The workshop received 10 submissions of which
7 papers were accepted for presentation, through a double-blind review process.
We would like to thank the Technical Program Committee members and the
reviewers for their time and dedication. We acknowledge support and guidance from
the dynamic Organizing Committee of ICVGIP 2016. In particular we appreciate
support received from Dr. Prithwijit Guha and Mathew Francis.
We hope that our efforts and the contributions presented in this collection will
encourage new ideas and further collaborative efforts between computer vision
researchers in industry and academia.

August 2017 Snehasis Mukherjee

Suvadip Mukherjee
Dipti Prasad Mukherjee
Jayanthi Sivaswamy
Suyash Awate
Srirangaraj Setlur
Anoop M. Namboodiri
Santanu Chaudhury
 Organization

Editors
Snehasis Mukherjee IIIT SriCity, India
Suvadip Mukherjee GE Global Research, India
Dipti Prasad Mukherjee Indian Statistical Institute, India
Jayanthi Sivaswamy IIIT Hyderabad, India
Suyash Awate IIT Bombay, India
Srirangaraj Setlur CEDAR, SUNY Buffalo, USA
Anoop M. Namboodiri IIIT Hyderabad, India
Santanu Chaudhury CSIR-CEERI Pilani, India

Technical Program Committee for WCVA

Aditya Nigam IIT Mandi, India
Afzal Godil National Institute of Standards and Technology, USA
Andrea Vaccari University of Virginia, USA
Balasubramanian Raman IIT Roorkee, India
Chetan Arora IIIT Delhi, India
Chiranjoy Chattopadhyay IIT Jodhpur, India
Devi Prasad Dogra IIT Bhubaneswar, India
Nilanjan Ray University of Alberta, Canada
Pratik Shah IIIT Vadodora, India
Rituparna Sarkar University of Virginia, USA
Sanmuganathan Raman IIT Gandhinagar, India
Shiv Ram Dubey IIIT SriCity, India
Soumitra Samanta University of Liverpool, UK
Swapna Agarwal TCS Innovation Lab, India
Angshuman Paul Indian Statistical Institute, India
Prithwijit Guha IIT Guwahati, India
Vineet Gandhi IIIT Hyderabad, India
P. Viswanath IIIT SriCity, India
Rajiv Ranjan Sahay IIT Kharagpur, India
Sanjoy Kumar Saha Jadavpur University, India

Technical Program Committee for MedImage

Jayanta Mukhopadhyay IIT Kharagpur, India
Niranjan Joshi BIRAC Regional entrepreneurship centre, Bangalore,
India
Ananda Shankar Jadavpur University, India
Chowdhury
VIII Organization

Gopal Joshi Noodle Analytics, India

Ajit Rajwade IIT Bombay, India
Anubha Gupta IIIT Delhi, India
Debdoot Sheet IIT Kharagpur, India
Bipul Das GE Global Research, India
Amit Kale Bosch Corporate Research, India
Arnav Bhavsar IIT Mandi, India

Technical Program Committee for DAR

Srinivasa Chakravarthy IIT Madras, India
Utpal Garain Indian Statistical Institute, India
Gaurav Harit IIT Jodhpur, India
C.V. Jawahar IIIT Hyderabad, India
Anoop M. Namboodiri IIIT Hyderabad, India
Atul Negi University of Hyderabad, India
Swapan Kumar Parui Indian Statistical Institute, India
A.G. Ramakrishnan IISc, India
Srirangaraj Setlur CEDAR, SUNY Buffalo, USA
 Contents

WCVA

A Novel Intelligent Multiple Watermarking Schemes for the Protection

of the Information Content of a Document Image . . . . . . . . . . . . . . . . . . . . 3
K.R. Chetan and S. Nirmala

Experimental Evaluation of 3D Kinect Face Database . . . . . . . . . . . . . . . . . 15

A.A. Gaonkar, M.D. Gad, N.T. Vetrekar, Vithal Shet Tilve, and R.S. Gad

Photometric Normalization Techniques for Extended Multi-spectral Face

Recognition: A Comparative Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
N.T. Vetrekar, R. Raghavendra, R.S. Gad, and G.M. Naik

Dictionary Based Approach for Facial Expression Recognition

from Static Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Krishan Sharma and Renu Rameshan

Vision Based Pose Estimation of Multiple Peg-in-Hole

for Robotic Assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Pitchandi Nagarajan, S. Saravana Perumaal, and B. Yogameena

A Spatio Temporal Texture Saliency Approach for Object

Detection in Videos. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
A. Sasithradevi, S. Mohamed Mansoor Roomi, and I. Sanofer

Simultaneous Reconstruction of Multiple Hand Shredded Content-Less

Pages Using Graph-Based Global Reassembly . . . . . . . . . . . . . . . . . . . . . . 75
K.S. Lalitha, Sukhendu Das, Arun Menon, and Koshy Varghese

Super Resolution Mapping of Trees for Urban Forest Monitoring

in Madurai City Using Remote Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
D. Synthiya Vinothini, B. Sathyabama, and S. Karthikeyan

Scale-Invariant Image Inpainting Using Gradient-Based

Image Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Mrinmoy Ghorai, Soumitra Samanta, and Bhabatosh Chanda

Recursive Structure from Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

M. Chebiyyam, S. Chaudhury, and I.N. Kar

A Hybrid Deep Architecture for Face Recognition in Real-Life Scenario . . . . 120

A. Sanyal, U. Bhattacharya, and S.K. Parui
X Contents

Brain Tumor Segmentation from Multimodal MR Images Using

Rough Sets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Rupsa Saha, Ashish Phophalia, and Suman K. Mitra

A Text Recognition Augmented Deep Learning Approach

for Logo Identification. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Moushumi Medhi, Shubham Sinha, and Rajiv Ranjan Sahay

High Frame Rate Real-Time Scene Change Detection System. . . . . . . . . . . . 157

Sanjay Singh, Ravi Saini, Sumeet Saurav, Pramod Tanwar,
Kota S. Raju, Anil K. Saini, Santanu Chaudhury, and Idaku Ishii

Painting Classification Using a Pre-trained Convolutional Neural Network . . . 168

Sugata Banerji and Atreyee Sinha

Eigen Domain Transformation for Soft-Margin Multiple Feature-Kernel

Learning for Surveillance Face Recognition . . . . . . . . . . . . . . . . . . . . . . . . 180
Samik Banerjee and Sukhendu Das

A Beta Distribution Based Novel Scheme for Detection of Changes

in Crowd Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
Soumyajit Pal, Sounak Mondal, Sanjoy Kumar Saha,
and Bhabatosh Chanda

Reconstruction of Sparse-View Tomography via Banded Matrices. . . . . . . . . 204

T. Prasad, P.U. Praveen Kumar, C.S. Sastry, and P.V. Jampana

DAR

SPODS: A Dataset of Color-Official Documents and Detection of Logo,

Stamp, and Signature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Amit Vijay Nandedkar, Jayanta Mukherjee, and Shamik Sural

Text and Non-text Separation in Scanned Color-Official Documents . . . . . . . 231

Amit Vijay Nandedkar, Jayanta Mukherjee, and Shamik Sural

Multi-font Telugu Text Recognition Using Hidden Markov Models

and Akshara Bi-grams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Koteswara Rao Devarapalli and Atul Negi

Anveshak - A Groundtruth Generation Tool for Foreground Regions

of Document Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
Soumyadeep Dey, Jayanta Mukherjee, Shamik Sural,
and Amit Vijay Nandedkar

Writer Identification for Handwritten Words . . . . . . . . . . . . . . . . . . . . . . . . 265

Shilpa Pandey and Gaurav Harit
 Contents XI

Kalanjiyam: Unconstrained Offline Tamil Handwritten Database . . . . . . . . . . 277

Faizal Hajamohideen and S. Noushath

Info-Graphics Retrieval: A Multi-kernel Distance Based Hashing Scheme . . . 288

Ritu Garg and Santanu Chaudhury

MedImage

Neovascularization Detection on Retinal Images . . . . . . . . . . . . . . . . . . . . . 301

Sudeshna Sil Kar, Santi P. Maity, and Seba Maity

Distribution Based EEG Baseline Classification . . . . . . . . . . . . . . . . . . . . . 314

Gopika Gopan K., Neelam Sinha, and Dinesh Babu J.

Texture Based Person Identification Using Dental Radiographs

and Photographs in Forensic Odontology . . . . . . . . . . . . . . . . . . . . . . . . . . 322
G. Jaffino, A. Banumathi, Ulaganathan Gurunathan, and J. Prabin Jose

Shearlet Based Medical Image Fusion Using Pulse-Coupled Neural

Network with Fuzzy Memberships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
Niladri Shekhar Mishra, Sudeb Das, and Amlan Chakrabarti

MR Imaging via Reduced Generalized Autocalibrating Partially Parallel

Acquisition Compressed Sensing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
Sheikh Rafiul Islam, Seba Maity, Santi P. Maity, and Ajoy Kumar Ray

Tracking of Retinal Microsurgery Tools Using Late Fusion of

Responses from Convolutional Neural Network over Pyramidally
Decomposed Frames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
Kaustuv Mishra, Rachana Sathish, and Debdoot Sheet

Cardiac Ultrasound Image Enhancement Using Tissue Selective Total

Variation Regularization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
Deepak Mishra, Santanu Chaudhury, Mukul Sarkar,
and Arvinder Singh Soin

Methods and System for Segmentation of Isolated Nuclei in Microscopic

Breast Fine Needle Aspiration Cytology Images . . . . . . . . . . . . . . . . . . . . . 380
Hrushikesh Garud, Sri Phani Krishna Karri, Debdoot Sheet,
Ashok Kumar Maity, Jyotirmoy Chatterjee, Manjunatha Mahadevappa,
and Ajoy Kumar Ray

Segmentation of Lumen and External Elastic Laminae in Intravascular

Ultrasound Images Using Ultrasonic Backscattering Physics Initialized
Multiscale Random Walks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
Debarghya China, Pabitra Mitra, and Debdoot Sheet

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405

WCVA
 A Novel Intelligent Multiple Watermarking
Schemes for the Protection of the Information
Content of a Document Image

K.R. Chetan(&) and S. Nirmala

Department of CSE, JNN College of Engineering, Shimoga, Karnataka, India

chetankr@jnnce.ac.in, nir_shiv_2002@yahoo.co.in

Abstract. Most of the past document image watermarking schemes focus on

providing same level of integrity and copyright protection for information
present in the source document image. However, in a document image the
information contents possess various levels of sensitivity. Each level of sensi-
tivity needs different type of protection and this demands multiple watermarking
techniques. In this paper, a novel intelligent multiple watermarking techniques
are proposed. The sensitivity of the information content of a block is based on
the homogeneity and relative energy contribution parameters. Appropriate
watermarking scheme is applied based on sensitivity classification of the block.
Experiments are conducted exhaustively on documents. Experimental results
reveal the accurate identification of the sensitivity of information content in the
block. The results reveal that multiple watermarking schemes has reduced the
amount of data to be embedded and consequently improved perceptual quality
of the watermarked image.

Keywords: Multiple watermarking
Intelligent watermarking
Fragile

watermarking
Robust watermarking
Integer wavelets
Contourlets

Gradient binarized blocks
GLCM

1 Introduction

Document images are used as proof for authentication and business transactions.
Traditionally digital watermarking has been used as a primary technique for copyright
protection and integrity management of document images [1–3]. The document image
consists of information with various levels of sensitivity. For instance, in a cheque
image, the signature and amount are dynamically changing information for each che-
que and thus possess highest level of sensitivity. The bank name, logo, cheque number
contain regeneratable information content and hence constitute lower level of sensi-
tivity. There also exists many empty areas in a cheque which can be classified as
insensitive areas. Each sensitivity level needs different type of protection. Therefore,
there is a need to use multiple watermarking techniques on the different areas of the
same document image. The multiple watermarking schemes have two fold objectives:
improve the perceptual quality of the watermarked image by reducing embedding
capacity; perform tamper detection and recovery with better accuracy.

© Springer International Publishing AG 2017

S. Mukherjee et al. (Eds.): ICVGIP 2016, LNCS 10481, pp. 3–14, 2017.
https://doi.org/10.1007/978-3-319-68124-5_1
4 K.R. Chetan and S. Nirmala

This paper is organized as follows: Sect. 2 provides a literature review of the

existing works in intelligent and multiple watermarking schemes. The proposed model
is explored in Sect. 3. Section 4 presents experimental results of the proposed multiple
watermarking scheme. Conclusions of the proposed work are summarized in the last
section.

2 Literature Review

Digital watermarking is classified as robust, fragile and semi-fragile based on the

robustness to incidental and intentional attacks [4]. A detailed survey of the works on
robust, fragile and semi-fragile watermarking techniques can be found in [5–10]. Most
of the past efforts on watermarking schemes apply single type of watermarking tech-
nique on the entire document image. Houmansadr et al. [11] proposed a watermarking
technique based on the entropy masking feature of the Human Visual System (HVS).
Kankanhalli and Ramakrishnan [12] developed a watermarking technique by embed-
ding just noticeable watermarks. Radharani et al. [13] designed a content based
watermarking scheme in which watermark is generated using Independent Component
Analysis (ICA) for each block of the input image. In [14–16], few works on the
segmentation of the image into objects using image statistics and subsequently
applying the robust watermarking schemes for each objects are described. Shieh et al.
[17] proposed the use of genetic algorithm (GA) [18] to compute the optimal frequency
bands for watermark embedding into a Discrete Cosine Transform (DCT) based
watermarking system, which can simultaneously improve security, robustness, and
image quality of the watermarked image. A novel idea was put forward in [19] to
embed multiple watermarks with different compression domains into the same source.
Lu et al. [19] developed an algorithm for embedding multiple watermarks into the
Vector Quantization (VQ) domain, as well as for hiding the secret keys associated with
the watermarks in the transform domain to enhance the robustness of the watermarked
image. Sheppard et al. [20] discussed the different ways of multiple watermarking like
rewatermarking, segmented watermarking and composite watermarking [20]. They
explored different attack scenarios [21, 22] and level of robustness that could be
provided by each category of multiple watermarking.
The literature reviews on the content based multiple watermarking techniques
reveals that most of the existing works lack intelligent classification of information
content of a document image based on sensitivity to the attacks. In the existing tech-
niques, authors attempted to apply multiple watermarks of the same type to each block
of the document image and it is not based on the appropriateness of the watermarking
for the information content present in the block. In addition, the existing schemes also
incur tradeoff between robustness and fragility of the watermarking multiple times.
These issues motivate towards an intelligent classification of the different areas of a
document image and application of different types of watermark schemes appropriate to
the sensitivity requirement of each area of the document image. In this paper, a new
model for intelligent multiple watermarking is designed that automatically computes
desired type of watermarking for each block of the document image.
 A Novel Intelligent Multiple Watermarking Schemes 5

3 Proposed Model

The proposed model for the novel intelligent multiple watermarking system consists of
two processes namely Embedding and Extraction. The Embedding process divides the
input document image into blocks and intelligently determines the type of water-
marking to be applied for each block. The watermarking algorithm depends on the
information content of the image. This is primarily available through the energy
component and hence luminance component in transformation is used. Further image is
converted back to color after watermarking to produce watermarked image. The
embedding technique depends on the type of watermarking. Robust watermarking is
implemented using integer wavelet embedding [23] and fragile watermarking is
accomplished using contourlet based embedding [24]. Extraction process is carried on
the blocks of the watermarked image. The result of the watermark extraction depends
on the type of the watermarking. The outcome of the robust watermark extraction on
the block of the watermarked image is content authentication of the block. The out-
come of fragile watermark extraction on the block of the watermarked image results in
tamper detection and recovery of information content in the block. The following
subsection explores the embedding process and extraction process in detail.

3.1 Multiple Watermark Embedding

The multiple watermark embedding process is shown in Fig. 1. It is an intelligent and
adaptive embedding scheme which depends on the information content of the

Fig. 1. Multiple watermark embedding process

6 K.R. Chetan and S. Nirmala

document image. Experiments have been conducted on all document images corpus to
analyze the effect of size of the block on accuracy in identification of the type of the
block and processing time. For each block, gradient binarized version of the infor-
mation content in the block is obtained. Further, the sensitivity level of each block and
type of watermarking required is found automatically. Subsequently, appropriate
watermark embedding algorithm is applied for each block.
The gradient binarized version of the information content in the block is computed
using the following algorithm:

In this algorithm, the values of the weights w1 and w2 is empirically set to 0.5. The
number of iterations required for termination of this algorithm depends on the distri-
bution of the information content in the block. The outcome of this algorithm is a
binary version of the block that gives segmentation of foreground and background
information contents in the block.
 A Novel Intelligent Multiple Watermarking Schemes 7

Experiments have been conducted on the document image corpus to decide on the
appropriate range to relative energy distribution and homogeneity values for deter-
mining the sensitivity levels of the blocks. The average REDb and HMb values for
different types of information content in these document images is calculated and
values are recorded in Table 1. It can be observed from the values in Table 1 that RED
values for blocks of document image containing dynamically changing information
content are in the range 0.7–0.85 and HM values lie between 0.29–0.50. Thus, sen-
sitivity level of the block with HM less than 0.5 and RED above 0.7 is set to 0.
Similarly, it can see in Table 1 that blocks of the document image containing pre-
printed information content has RED values above 0.3 and HM values in between 0.5
to 0.85. Therefore, sensitivity level of these blocks is set to 1. For all the other blocks,
sensitivity level is set to 2.
The type of watermarking used depends on the sensitivity levels of the information
content in the block. Highly sensitive blocks are protected using fragile watermarking
technique. In this paper an effective fragile watermarking technique based on con-
tourlets [24] is used. Partially sensitive blocks are protected using robust watermarking
technique [23]. The size of the block is decided based on two factors: effectiveness in
the identification of the sensitivity of the block and processing time for identification.
8 K.R. Chetan and S. Nirmala

Table 1. Computation of RED and HM values for different classes of document images in the
corpus
Document Preprinted Dynamically changing
image class information information
RED HM RED HM
Cheques 0.33 0.67 0.72 0.50
Bills 0.36 0.82 0.79 0.35
Identity cards 0.62 0.85 0.76 0.39
Marks cards 0.30 0.79 0.85 0.29
Certificates 0.54 0.51 0.70 0.36

Experiments have been conducted exhaustively on all the document images to measure
the impact of size of the block against accuracy in identifying sensitivity level of the
block. The average number of blocks expected for each sensitivity level and number of
blocks being accurately identified is recorded in Table 2. It can be observed from
average accuracy in identification values that the blocks of lesser size exhibits higher
accuracy.

Table 2. Impact of size of the blocks of a document image on accuracy in identification of its
sensitivity level and processing time for identification
Block size Type-0 Type-1 Type-2 IA (in %) Processing time (in secs)
EB IB EB IB EB IB
32  32 348 339 210 204 466 466 98.53 102.12
64  364 83 75 57 52 116 116 94.92 78.7
128  3128 22 20 17 15 25 25 93.75 32.11
256  3256 5 4 4 3 7 7 87.50 21.16

Where, EB-Expected no. of blocks evaluated manually by an expert, IB-Identified

No. of blocks from the proposed approach, IA-Identification Accuracy of a block,
which is calculated as the ratio of sum of IB of all block types over sum of EB of all the
block types. Considering the values of the accuracy in identification of sensitivity level
of a block and processing time incurred for identification, size of the block is set to
128  128.

3.2 Multiple Watermark Extraction

Multiple watermark extraction process has two fold objectives based on the of
watermark extraction process involved. Robust watermark extraction aims at content
authentication of the block. This is implemented using robust watermarking scheme
[23]. Fragile watermark extraction involves tamper detection and recovery of the
information content of a document image. The fragile watermark extraction is per-
formed using contourlets [24]. Multiple watermark extraction has similar steps as in
multiple watermark embedding process discussed in Sect. 3.1 until the identification of
the type of the gradient binarized block. Subsequently, the type of the block extracted
 A Novel Intelligent Multiple Watermarking Schemes 9

and generated is compared and if there is a mismatch, the corresponding block of the
document image is declared “inauthentic”. However, if there is a match, then water-
mark extraction is carried out based on the type of the block. The extracted and
generated watermarks are compared for similarity using Feature Similarity Index [24]
and based on the comparison, the tamper detection of the block is decided. If the block
is tampered, recovery of information content is made by extracting watermark
embedded at robust locations [24]. During robust watermark extraction, the watermark
is extracted from the LL-band of the integer wavelet transformation performed on the
block of a document image. The extracted watermark is decoded using binary block
coding technique [23]. The decoded watermark is compared with original watermark
and decision of content authentication of the block is performed [23].

4 Results

We have created a corpus of document images. All the images in the corpus are
scanned document images. The classes of document image corpus considered are
Cheques, Bills, Identity Cards, Marks cards and Certificates. Each class of document
image consists of 30 images. The results of the identification of type of the blocks of a
sample document image in the corpus are shown in Fig. 2.

Fig. 2. Results of identification of the type of the block of a sample Cheque image

It can be observed in Fig. 2, that there are three types of blocks in the sample
Cheque image. The blocks with dashed border are Type-0 blocks. They are highly
sensitive blocks containing large variations in the information content and distribution
of the information. The blocks with dotted border are Type-1 blocks i.e. partially
sensitive blocks which contain preprinted information. They have moderate homo-
geneity in distribution of the information content. Remaining type of blocks in the
document image are the insensitive blocks (Type-2) which contain less energy and
10 K.R. Chetan and S. Nirmala

higher homogeneity of information. We have tested the accuracy of the identification

for all the classes of document images in the corpus.
Once the blocks are identified, appropriate type of watermarking has been applied
based on the type of the block to obtain watermarked image. Subsequently multiple
watermark extraction has been applied on the watermarked image and incidental and
intentional attacks have been applied on the watermarked image. The results of mul-
tiple watermark embedding and extraction are shown in Fig. 3. Figure 3 shows that
watermarked image is perceptually similar to source document image in the corpus. An
example of incidental attack on partially sensitive block and intentional attack on a
highly sensitive block of the watermarked image is demonstrated in Fig. 3. Further, one
could also observe there is great degree of accuracy in tamper detection and recovery of
the highly sensitive block.

Fig. 3. Results of proposed multiple watermarking system (a) source document image
(b) watermarked image (c) original watermark for robust watermarking (d) zoomed up Partially
sensitive block with salt and pepper noise attack (e) zoomed up Highly sensitive tampered block
(f) extracted robust watermark (g) tamper detection results (h) tamper recovery result

5 Analysis

The performance of the proposed watermarking system is measured in terms of the

following parameters: (i) Performance analysis using Peak Signal to Noise Ratio
(PSNR) (ii) Robustness Analysis using Normalized Correlation Coefficient (NCC) (iii)
Fragility Analysis using accuracy of Tamper detection and recovery.

5.1 Performance Analysis

The performance of the proposed multiple watermarking scheme is evaluated in terms
of PSNR. The perceptual quality of the watermarked image of size NXN is measured
 A Novel Intelligent Multiple Watermarking Schemes 11

using Peak Signal to Noise Ratio (PSNR) [25]. A graph of PSNR values is depicted in
Fig. 4 for different classes of the document images. The graph shown in Fig. 4 reveals
that PSNR values of the multiple watermarking schemes are better than robust and
fragile watermarking schemes when applied separately. This increase in PSNR and
subsequently the perceptual quality of the watermarked image is due to the fact that all
the blocks of the document image are not watermarked. The quantity of the watermark
to be embedded depends on the type of the block. Hence, the noise induced due to
watermarking is reduced to some extent and this result in the better fidelity of the
watermarked image.

Fig. 4. Effect of watermarking schemes on PSNR values of different classes of document

images in the corpus

5.2 Robustness Analysis

The robustness of the proposed multiple watermarking scheme is tested by applying
various attacks such as horizontal cropping, vertical cropping, resizing, noise and JPEG
compression on all the document images in the corpus. The degree of robustness
obtained is evaluated in terms of NCC: [26]. The NCC values obtained by the appli-
cation of proposed watermarking scheme only on partially sensitive blocks and robust
watermarking scheme applied on the entire document image is recorded in Table 3.
The NCC values in Table 3 show that there is a slight improvement in the robustness of
the watermarked image. The increase in robustness is due to the localization of
robustness to the blocks that are partially sensitive.

Table 3. Average NCC values for different incidental attacks

Incidental attack Existing Robust Proposed multiple
watermarking scheme [23] watermarking scheme
Salt and pepper noise 0.93 0.96
Cropping 0.97 0.97
Resizing 0.94 0.95
JPEG compression 0.94 0.96
12 K.R. Chetan and S. Nirmala

5.3 Fragility Analysis

The fragility capability of any watermarking scheme is evaluated in terms of accuracy
of tamper detection and tamper recovery parameters. Accuracy of tamper detection is
evaluated as follows:
Pn
i¼1 ðtai  tdi Þ
TDA ¼ 1  ð7Þ
n
where, n – total number of bits in the fragile watermarked blocks, ta – tampered bit, td –
tamper detection bit. The average values of TDA and TRA are computed for all doc-
ument images in the corpus under different intentional attacks for proposed fragile
watermarking scheme and contourlet based scheme [24] separately. These values are
tabulated in Table 4. It can be observed that proposed multiple watermarking schemes
has a slight improvement capability in detection and recovering from tampering of
information content of document image.

Table 4. Average TDA and TRA values for different intentional attacks
Intentional attacks Existing Fragile water Proposed multiple
marking scheme [24] watermarking scheme
TDA TRA TDA TRA
Insertion 0.9 0.87 0.91 0.90
Deletion 0.92 0.91 0.92 0.92
Modification 0.87 0.87 0.90 0.89

6 Conclusions

A novel intelligent multiple watermarking schemes are proposed in this paper. The
blocks of a document image have been automatically classified into various sensitivity
levels with greater accuracy. The performance analysis of the proposed approach
reveals improvement in the perceptual quality of the watermarked image. The proposed
scheme also outperforms the existing methods [23, 24] in providing robustness, tamper
detection and recovery capabilities. Improvement on the accuracy of identification of
type of block is taken up as future work of the current study.

References
1. Wu, M., Liu, B.: Watermarking for image authentication. In: Proceedings of the IEEE
International Conference on Image Processing, pp. 437–441 (1998)
2. Cox, I., Miller, M., Bloom, J., Fridrich, J., Kalker, T.: Digital Watermarking And
Steganography. Morgan Kaufmann Publishers Inc., San Francisco (2007)
3. Hartung, F., Kutter, M.: Mutimedia Watermarking Techniques. Proc. IEEE 87(7), 1079–
1107 (2002)
 A Novel Intelligent Multiple Watermarking Schemes 13

4. Potdar, V.M., Han, S., Chang, E.: A survey of digital image watermarking techniques. In:
3rd IEEE International Conference on Industrial Informatics, pp. 709–716 (2005). doi:10.
1109/Indin.2005.1560462
5. Mirza, H., Thai, H., Nakao, Z.: Color image watermarking and self-recovery based on
independent component analysis. In: Rutkowski, L., Tadeusiewicz, R., Zadeh, L.A., Zurada,
J.M. (eds.) ICAISC 2008. LNCS, vol. 5097, pp. 839–849. Springer, Heidelberg (2008).
doi:10.1007/978-3-540-69731-2_80
6. Wang, M.S., Chen, W.C.: A majority-voting based watermarking scheme for color image
tamper detection and recovery. Comput. Stand. Interfaces 29, 561–571 (2007)
7. Bas, P., Chassery, J.M., Macq, B.: Geometrically invariant watermarking using feature
points. IEEE Trans. Image Process. 11(9), 1014–1028 (2002)
8. Qi, W., Li, X., Yang, B., Cheng, D.: Document watermarking scheme for information
tracking. J. Commun. 29(10), 183–190 (2008)
9. Dawei, Z., Guanrong, C., Wenbo, L.: A chaos-based robust wavelet-domain watermarking
algorithm. Chaos, Solitons Fractals 22(1), 47–54 (2004)
10. Schirripa, G., Simonetti, C., Cozzella, L.: Fragile digital watermarking by synthetic
holograms. In: Proceedings of the European Symposium on Optics/Fotonics in Security &
Defence, London, pp. 173–182 (2004)
11. Houmansadr, A., et al.: Robust content-based video watermarking exploiting motion entropy
masking effect. In: Proceedings of the International Conference on Signal Processing and
Multimedia Applications, pp. 252–259 (2006)
12. Kankanhalli, M.S., Ramakrishnan, K.R.: Adaptive visible watermarking of images. In: IEEE
International Conference on Multimedia Computing and Systems, vol. 1, pp. 568–573
(1999)
13. Radharani, S., et al.: A study on watermarking schemes for image authentication. Int.
J. Comput. Appl. (0975 – 8887) 2(4), 24–32 (2010)
14. Kay, S., Izquierdo, E.: Robust content based image watermarking. In: Proceedings of the
Workshop on Image Analysis for Multimedia Interactive Services (2001)
15. Kim, M.-A., Lee, W.-H.: A content-based fragile watermarking scheme for image
authentication. In: Chi, C.-H., Lam, K.-Y. (eds.) AWCC 2004. LNCS, vol. 3309,
pp. 258–265. Springer, Heidelberg (2004). doi:10.1007/978-3-540-30483-8_31
16. Habib, M., Sarhan, S., Rajab, L.: A Robust-Fragile dual watermarking system in the DCT
domain. In: Khosla, R., Howlett, R.J., Jain, L.C. (eds.) KES 2005. LNCS, vol. 3682,
pp. 548–553. Springer, Heidelberg (2005). doi:10.1007/11552451_74
17. Shieh, C.-S., et al.: Genetic watermarking based on transform-domain techniques. J. Pattern
Recogn. 37, 555–565 (2004)
18. Goldberg, D.E.: Genetic Algorithms in Search Optimization and Machine Learning.
Addison-Wesley, Reading (1992)
19. Lu, Z.-M., Xu, D.-G., Sun, S.-H.: Multipurpose image watermarking algorithm based on
multistage vector quantization. IEEE Trans. Image Process. 14(6), 822–831 (2005). doi:10.
1109/Tip.2005.847324
20. Sheppard, N.P., Safavi-Naini, R., Ogunbona, P.: On multiple watermarking. In: Dittmann, J.,
Nahrstedt, K., Wohlmacher, D. (eds.) Multimedia and Security: New Challenges Workshop,
p. 38871 (2001)
21. Voloshynovskiy, S., Pereira, S., Pun, T., Eggers, J.J., Su, J.K.: Attacks on digital
watermarks: classification, estimation based attacks, and benchmarks. IEEE Commun. Mag.
39(8), 118–126 (2001)
22. Wang, S., Zhang, X.: Watermarking scheme capable of resisting sensitivity attack. IEEE
Signal Process. Lett. 14(2), 125–128 (2007)
14 K.R. Chetan and S. Nirmala

23. Chetan, K.R., Nirmala, S.: An efficient and secure robust watermarking scheme for
document images using integer wavelets and block coding of binary watermarks. J. Inf.
Secur. Appl. 24–25, 13–24 (2015)
24. Chetan, K.R., Nirmala, S.: A novel fragile watermarking scheme based on contourlets for
effective tamper detection, localization and recovery of handwritten document images. IEEE
Signal Process. Lett. (Communicated)
25. Aggarwal, E.D.: An efficient watermarking algorithm to improve payload and robustness
without affecting image perceptual quality. J. Comput. 2(4) (2010). ISSN 2151-9617
26. Zhu, X., et al.: Normalized correlation-based quantization modulation for robust water-
marking. IEEE Trans. Multimed. 16(7), 1888–1904 (2014)
 Experimental Evaluation of 3D Kinect Face
Database

A.A. Gaonkar1(&), M.D. Gad2, N.T. Vetrekar1, Vithal Shet Tilve3,

and R.S. Gad1
1
Department of Electronics, Goa University, Taleigao Plateau, Goa, India
{elect.aagaonkar,elect.ntvetrekar,rsgad}@unigoa.ac.in
2
Goa Engineering College, Farmagudi, Goa, India
miteshgad92@gmail.com
3
School of Earth and Space Exploration Arizona State University, Tempe, USA
tilvi@asu.edu

Abstract. 3D face recognition has gain a paramount importance over 2D due to

its potential to address the limitations of 2D face recognition against the vari-
ation in facial poses, angles, occlusions etc. Research in 3D face recognition has
accelerated in recent years due to the development of low cost 3D Kinect camera
sensor. This has leads to the development of few RGB-D database across the
world. Here in this paper we introduce the base results of our 3D facial database
(GU-RGBD database) comprising variation in pose (0°, 45°, 90°, −45°, −90°),
expression (smile, eyes closed), occlusion (half face covered with paper) and
illumination variation using Kinect. We present a proposed noise removal
non-linear interpolation filter for the patches present in the depth images. The
results were obtained on three face recognition algorithms and fusion at
matching score level for recognition and verification rate. The obtained results
indicated that the performance with our proposed filter shows improvement over
pose with score level fusion using sum rule.

1 Introduction

Facial expressions, poses and variations have attracted the research world since long, as
it is easily obtainable and convenient biometric trait as compared to the iris, voice, gait
etc. 2D facial images has well defined roots in the world of biometric research due to
low cost of its acquisition system and wide availability [1]. But 2D face recognition
system faces its limitations when it comes to mostly illumination and pose variation [2].
In order to overcome these short comings of 2D recognition, 3D recognition system
captured the market as security concern has increased from local to the defense level.
Research in 3D biometric was an expensive task as the expense of system requirement
for acquiring 3D images was very high and time consuming [3] until the development
of efficient, low-cost RGB-D Kinect camera. This system provides 2D RGB image as
well as depth information i.e. distance from each pixel to the sensor [4].
Images captured by Kinect camera has low resolution and noisy yet it has more
spatial information compared to 2D in form of depth which is a robust inherent
property associated for 3D face recognition against uncontrolled environment.

© Springer International Publishing AG 2017

S. Mukherjee et al. (Eds.): ICVGIP 2016, LNCS 10481, pp. 15–26, 2017.
https://doi.org/10.1007/978-3-319-68124-5_2
16 A.A. Gaonkar et al.

Ekenel et al. obtained 3-D face recognition approach using the discrete cosine trans-
form (DCT) which is a local appearance-based model at feature level [5]. Huynh et al.
has proposed a new LBP based descriptor namely Gradient-LBP (G-LBP) for gender
recognition task on EURECOM and Texas database [6]. Vezzetti et al. proposed a new
3D face recognition algorithm, whose framework based on extracting facial landmarks
using the geometrical properties of facial shape [7]. Min et al. have generated a 3D
database based on the Kinect sensor having 52 subjects over two sessions for 2D, 2.5D,
3D and video. Here recognition rates are calculated for 2D, 2.5D, and 3D-based face
data using standard face recognition techniques like PCA, LBP, SIFT, LGBP, ICP, and
TPS and also RGB and Depth images were fused using score-level fusion [8]. Ajmera
et al. has computed CRR based on modified SURF descriptors and image enhancement
techniques and filters like adaptive histogram equalization, NLM filter etc. for their
internal database and has compared it with Eurecom and Curtin face database and also
has performed scored level fusion [9]. Hg et al. had proposed RGB-D Face dataset
(VAP database) of 31 subjects containing 1581 images and have developed a face
detection protocol using curvature analysis technique and tested for VAP database [10].
Goswami et al. had generated IIIT-D RGB-D face database of 106 subjects with
multiple Images per subject. Authors has also proposed an algorithm for 3D face
recognition which involves computation of entropy map and visual saliency map
followed by HOG descriptor for feature extraction and finally the use of Random
Decision Forest (RDF) classifier for establishing identity. The algorithm was tested for
IIIT-D and Eurecom Database [4, 11]. Table 1 is giving the brief idea about existing
Kinect base 3d facial databases.

Table 1. Existing 3D facial databases

Database No. of Variations
subjects Angles/poses Occlusion Expressions Illumination
Eurecom [8] 52 Neutral face, Paper and Smile, Single pose
right, left hand on mouth open
face,
sunglasses
VAP [10] 31 Combination of – Smile, sad, –
17 vertical and yawn,
horizontal face anger
poses
Curtin face 100 Various poses – Various Yes
[3] expressions
IIIT-D [11] 106 Various poses – Various Yes
expressions
IIT-K [9] 100 0°, 15°, 30°, 45°, – Yes
60°, 75°, 90°
VT-KFER 32 Frontal, right, left – 6 facial –
[20] expressions
GU-RGBD 64 0°, 45°, 90°, Paper on Smile, eyes Full
−45°, −90° face close session 2
 Experimental Evaluation of 3D Kinect Face Database 17

Mao et al. has work on identification and filling of expansion holes. Here the holes
are identified based on depth histogram and the filling of holes is done using linear
interpolation and graph based interpolation method [13]. Solh and AlRegib has pro-
posed two approaches for dis-occlusion removal in Depth Image-Based Rendering
(DIBR): hierarchical hole-filling (HHF) and depth adaptive hierarchical hole-filling.
The said approaches follows pyramid like approach from lower resolution estimate of
3D wrapped image to estimate the said hole pixels [14]. Wang et al. have propose a
hole filling algorithm to improve image quality of DIBR. Here to determine the order of
hole filling the depth information is added to the priority calculation function. Next
when searching for the best matching block the gradient information is used as aux-
iliary information [15, 16]. Feng et al. has proposed an adaptive background biased
depth map hole-filling method [17]. Based on this literature survey we are proposing
simple weighted average nonlinear interpolation hole/patch removal algorithm for the
3D database generated at our laboratory. We are presenting our GU-RGBD facial
database having variation in pose, expressions and occlusion; collected over two ses-
sions for 64 subjects. This database will be made available in public domain for
research purpose. Extensive experimental evaluation is performed for the State of the
Art face recognition algorithm in the form of identification and verification rate.
The rest of the paper is organized as follows: Sect. 2 describes the GU-RGBD data
generation setup and protocol. Section 3 is giving the detail explanation of the pro-
posed filter. Section 4 explains the experimental evaluation protocol for database.
Section 5 comprises of results and discussion and the final conclusion is given in
Sect. 6.

2 3D Database Generation

2.1 3D Imaging Setup

3D biometric imaging laboratory having controlled and uncontrolled environmental
conditions was setup at our work place. This setup was comprised of RGBD camera,
light sources, and a computer system. RGBD images were captured using Xbox 360
Kinect depth camera from Microsoft, which consists of two parts, RGB camera to
capture 2D image information and depth sensor which acquires depth information.
The depth sensor has an infrared projector combined with a monochrome CMOS
sensor, which captures 3rd dimension (distance between subject and sensor) i.e. depth.
The sensor placement was at the height of 1.5 mts from the floor and approximately at
the distance of 1.25 mts from the subject. To set the controlled condition two QTH
light sources of 600 W were kept at an angle 45° normal to the subject position. In
order to maintain the uniform background and equal illumination on all sides white
muslin cloth backdrop was mounted behind the subject. The direct projection of light
on the subject was avoided with white muslin cloth umbrellas in front of light sources.
To set uncontrolled environmental condition the subject was exposed to the ambient
light by opening the windows while capturing the images.
18 A.A. Gaonkar et al.

2.2 3D Image Acquisition Protocol

The image acquisition was performed after proper calibration of camera to confirm the
constant parameters throughout the experiment. Here the highest resolution for Kinect
color sensor (1280
960) and Kinect depth sensor (640
480) was selected. Data-
base was collected in two sessions (controlled and un-controlled) for the students and
staff of our organization. The image capturing protocol was designed as shown in
Fig. 1. We have introduced eight variations per individual in the image acquisition
process, having variation in pose (−90°, −45°, 0°, +45°, +90°), expressions (smile,
eyes closed) and occlusion (paper was covering half part of the face).

Fig. 1. 3D image acquisition protocol

For every subject, eight RGB and eight depth images were captured to form total of
16 in every session. The session 1 consists 16 images captured in controlled envi-
ronmental condition and session 2 consist of 16 images captured in uncontrolled
environmental condition by keeping window open for ambient light to enter the lab and
the QTH sources were off. Thus total of 32 images captured per subject. The
GU-RGBD database has the enrollment of 64 subjects out of which 49 are males and
15 are females from various age group. The size of database is 64(subjects) * 32
(images per subject) = 2048 images. The sample images of database are shown in
Fig. 2.

Fig. 2. 3D image acquisition protocol

 Experimental Evaluation of 3D Kinect Face Database 19

3 Proposed Patch Removal Filter

The images captured by Kinect are noisy and inaccurate [8]. The presence of patches
(zero value pixels present on the depth face) degrade the captured information and thus
affects the recognition rate and hence it has to be enhanced in the pre-processing stage.
We are using the interpolation method to replace the patch with the neighboring pixels.
There are various scenarios of the characteristics of patch like localized position on
facial triangle, size of patch and patch position on the boundaries. Hence sometimes
surrounding information if not available in such scenarios especially when patch is
placed at boundary positions. Hence, there is a need to allocate the weightage for the
populace of surrounding pixels in the form of some constants i.e. a1, a2. The proposed
filter acts as the solution for patches present on depth faces.
The depth images of M
N dimensions usually having the 4:3 aspect ratio are
extended by adding M/4 dummy rows and column pixels respectively for the higher
dimensions of aspect ratio as shown in Fig. 3(a). This is extended so as to avoid the
occurrence of computational errors for the pixels at the outer boundary as window of
filter expands. Then ‘NaN’ values are assigned to the dummy rows and column pixels
so as to avoid filling of false information for patches in vicinity of boundary. The
proposed filter scan image by initially setting a l
l kernel i.e. ‘u(l,l)’ to locate zero
value pixel (which is a patch) to be replaced.

Fig. 3. Filter implementation: (a) schematic view of filter, (b)–(d) patches of different sizes
Another random document with
no related content on Scribd:
however, the more he washed in the hot tears, the redder grew his
clothes, until he was just the colour of the scarlet bean blossom.
“You told me a story,” said Gillydrop to the giant when he saw how
red he was getting.
“I know I did,” said Dunderhead, drying his eyes, for he had now
wept enough, and was growing hungry; “but if I hadn’t told you a
story, I wouldn’t have got any supper. You’ll never be green again, so
don’t trouble your head. I’m going to get some wood to cook these
nice fat children.”
On hearing this, Teddy and Tilly roared like bulls, and Gillydrop
roared too, for he was afraid he would never be able to go back to
Faeryland in his red clothes; but the giant only laughed at them, and
went out to light a fire under his big kettle.
Gillydrop was naturally very cross with the giant for having deceived
him, and determined to punish him for having done so. Bringing the
two children to Dunderhead for his supper could not be the kindly
deed he had to do, or else he would have turned green again; so
Gillydrop made up his mind to take Teddy and Tilly back to earth,
and thus leave Dunderhead without his supper. While he was thus
making up his mind, seated at one end of the table, the two children,
seated at the other end, were crying bitterly at the plight in which
they now found themselves, for it certainly is not a nice thing to be
boiled for an ogre’s supper.
“Poor mother!” wailed Tilly, weeping; “she’ll miss us so much.”
“I don’t know if she will,” replied Teddy dolefully; “we’ve always been
so naughty, I daresay she’ll be glad we’ve gone.”
“Oh no, she won’t,” said Tilly, nodding her head; “she loves us too
much for that; but if we could get back I’d be so good.”
“And so would I,” cried Teddy; and then they both wept again, while
Gillydrop, seeing their tears, wept also out of sheer sympathy.
“Perhaps the giant will only eat one of us,” said Tilly after a pause;
“so while one of us is boiling, the other must run away and go back
to comfort mother.”
“Who will be boiled?” asked Teddy sadly. “Will you, Tilly?”
“I don’t like being boiled,” answered Tilly, with a shudder. “I’m sure it
isn’t nice.”
“Well, I don’t like being boiled either,” observed Teddy. “Suppose we
draw lots who is to run away.”
“Yes, that would be fair,” said Tilly, drying her eyes; “and the one who
wins must go back to cheer mother.”
Gillydrop was quite sorry now that he had brought them for
Dunderhead’s supper, when he heard how they regretted their
mother; so he made up his mind to save them.
“You shall neither of you be boiled,” he said, walking up to them
across the table, which was like a large plain. “I will take you back to
your mother.”
“But how?” asked Teddy and Tilly, both together. “We cannot go back
across the sea alone.”
“Oh yes, you can,” replied the Red Elf. “I brought you here, and can
send you back; that is, if I only had a leaf.”
“Here is one,” cried Tilly eagerly, pulling a faded leaf out of her
pocket. “I picked it up in the wood to-day, it had such pretty red and
yellow colours.”
“Oh, that will do for a boat,” said Gillydrop joyfully.
“But it’s so small,” objected Teddy.
“I’ll make it large enough,” said the elf. “You’ll see.”
“But how can we go on without sails or oars?” said Tilly timidly.
“You don’t need any,” rejoined Gillydrop, laughing; “you know every
tree has power to draw back its own leaves. The boat we came in
was a leaf, and, as soon as it was launched on the air, it went
straight back to the tree in the Country of the Giants upon which it
had grown; and as this leaf comes from a tree on earth, it will go
straight back to its tree.”
“Then we can get home,” cried Tilly, clapping her hands, “for the tree
isn’t far from mother’s cottage.”
“Mind, you are never to be naughty again,” said Gillydrop solemnly.
“Oh, no, no!” cried both children.
“And be very, very good to your mother.”
“Yes, yes! We’ll be very good.”
“Then go down to the beach by the path,” said Gillydrop, spreading
his wings. “I’ll fly down and get the boat ready; be quick, or the giant
will return.”
Then he flew away through the open window, and Teddy scrambled
down the steep path, followed by Tilly, both of them in a great fright
lest the giant should catch sight of them and pop them into his big
kettle. When they reached the beach, they found Gillydrop had
launched the leaf, which had now been transformed into a beautiful
red and yellow coloured boat.
“Good-bye,” said Gillydrop, as soon as they were comfortably seated
in the boat. “I’m sorry I brought you here, but it will do you no harm,
as it will teach you to be good. Mind you don’t quarrel in the boat—if
you do, the leaf will vanish, and you’ll sink for ever in the black
waves.”
“Oh, we’ll be very, very good,” promised both the children eagerly,
and then Gillydrop gave the boat a push, so that it moved rapidly
away from the land, leaving him seated on the beach, a lonely little
red figure.
Teddy and Tilly were rather afraid at finding themselves alone in the
darkness, but they kissed one another, and fell asleep, while the
leaf-boat sailed rapidly over the Sea of Darkness towards its parent
tree. When the children awoke, they found themselves lying on the
ground under the tree, and there above them was their red and
yellow boat, hanging, a red and yellow leaf, on a high bough.
“Now we’ll go home,” cried Tilly, jumping up; “now we’ll go home to
mother.”
“And be very good,” said Teddy, also rising.
“Yes; very, very good,” replied Tilly. And then, taking one another’s
hands, they ran home to their cottage through the dark forest.
Dame Alice, who thought they had lost themselves in the wood, was
very glad to see them, and, after she had kissed them, gave them a
good supper of bread and milk, which they enjoyed very much, for
you see they were very hungry with the long journey.
They told Dame Alice all their adventures, and she was very glad
they had gone to the Giants’ Country, for she guessed, like the wise
mother she was, that this was the lesson the faeries had foretold.
Ever afterwards, Teddy and Tilly were good children; there never
were two such good children, because they thought, if they were not
good, they would be taken back to the Giants’ Country and boiled for
an ogre’s supper. But after a time they liked to do good actions
because they found it pleasant, and Dame Alice was so pleased with
their behaviour that she made a rhyme about them, which soon
passed into a proverb:
“The magic power of a faery
Cures a child when quite contrary.”
 III.

HOW THE RED ELF RETURNED TO FAERYLAND.

When Gillydrop saw the magic boat disappear into the darkness of
the sea, he thought that, now he had done one kindly deed, his
clothes would change from red to green, and he would be able to
return to his dear Faeryland. But nothing of the sort occurred, and
the poor elf began to cry again, thinking he was lost for ever, but this
time his tears were not red, which was a good sign, although he did
not know it.
Very soon he heard Dunderhead roaring for the loss of his supper,
so, drying his eyes, he flew back again to the hall of the castle, to
see what the giant was doing. He found a great fire was lighted, over
which was suspended a great kettle filled with water, which was now
boiling hot. Dunderhead was searching everywhere for the children,
and when he saw Gillydrop he shook his great fist at him.
“Where’s my supper, you red rag?” he roared fiercely.
“Your supper has gone back to earth,” replied Gillydrop angrily, for no
one likes to be called a red rag. “You told me a story, so I thought I’d
punish you.”
“Oh, did you?” bellowed Dunderhead, in a rage. “Then I’ll punish you
also for spoiling my supper.” And before Gillydrop could fly away, he
caught him in his great hand and popped him into the boiling water.
Oh, it was terribly hot, and Gillydrop thought it was all over with him;
but, being a Faery, he could not be killed, as the foolish giant might
have known. He sank down, down, right to the bottom of the great
kettle, and then arose once more to the top. As soon as he found his
head above water, he sprang out of the kettle and flew away high
above the head of Dunderhead, who could only shake his fist at him.
To his delight and surprise, Gillydrop found his clothes had all
changed from red to green, and instead of being dressed in crimson,
his suit was now of a beautiful emerald colour. He was so delighted
that he flew down on to the floor of the hall, and began to dance and
sing, while the giant joined in as he tried to catch him; so that they
had quite a duet.
Gillydrop. Now I’m gay instead of sad,
For I’m good instead of bad:
Dreadful lessons I have had.
Giant. I will catch and beat you!

Gillydrop. Tho’ a naughty elf I’ve been,

Now my clothes are nice and clean:
I dance once more a faery green.
Giant. I will catch and eat you!

But you see he could not do that, because Gillydrop was too quick
for him, and flew round the hall, laughing at Dunderhead, who roared
with anger. Then the elf flew out on to the terrace which overlooked
the Sea of Darkness, followed by the giant. Gillydrop flew down on to
the beach to escape the ogre, and Dunderhead tried to follow; but,
as he could not fly, he fell right into the Sea of Darkness. Dear me!
what a terrible splash he made! The waves arose as high as the
castle walls, but then they settled down again over Dunderhead, who
was suffocated in the black billows. He was the very last of the
giants, and now his bones lie white and gleaming in the depths of the
Sea of Darkness, where nobody will ever find them—nor do I think
any one would trouble to look for them.
As for Gillydrop, now that Dunderhead was dead, he flew away
across the dreary plain towards Faeryland, and soon arrived at the
borders of the sullen grey sea which still rolled under the pale light of
the moon. Gillydrop was not a bit afraid now, because his clothes
were green once more, and he had performed one kindly deed; so
he sat down on the seashore and sang this song:
“When from Faeryland I fled,
All my nice clothes turned to red;
Now in emerald suit I stand—
Take me back to Faeryland.”
And as he sang the grey ocean faded away, and in its place he saw
the green trees of the faery forest, waving their branches in the silver
moonlight. Only a bright sparkling stream now flowed between
Gillydrop and Faeryland; so, spreading his silver and blue wings, he
flew across the water, singing gaily:
“Thanks, dear Oberon. At last
All my naughtiness is past;
Home I come without a stain,
And will never roam again.”

So at last Gillydrop got back to Faeryland after all his trials, and ever
afterwards was one of the most contented elves ever known. You
may be sure he never wanted to see the Country of the Giants
again, and whatever King Oberon said he did willingly, because he
knew it must be right.
He was quite a hero among the faeries, and had the honour of telling
all his adventures to King Oberon himself, which he did so nicely that
the King gave him a title, and ever afterwards he was called “Sir
Gillydrop the Fearless.”
 SHADOWLAND

IT was Christmas Eve, and the snow, falling heavily over a great city,
was trying to hide with its beautiful white robe all the black, ugly
houses and the narrow, muddy streets. The gas lamps stood up
proudly, each on its tall post, and cast their yellow light on the
crowds of people hurrying along with their arms filled with many
lovely presents for good children.
“They are poor things,” said the gas lamps scornfully. “If we did not
shed our light upon them, they would be lost in the streets.”
“Ah, but the people you despise made you,” cried the church bells,
which were calling the people to prayer. “They made you—they
made you, and gave you your beautiful yellow crowns.”
But the street lamps said nothing, because they could not deny what
the church bells said, and instead of acknowledging that they owed
all their beauty to the people they despised, remained obstinately
silent.
Near one of these lamp-posts, at the end of a street, stood a
ragged boy, who shivered dreadfully in his old clothes, and stamped
about to keep himself warm. The boy’s name was Tom, and he was
a crossing-sweeper, as could be seen by his well-worn broom. He
was very cold and very hungry, for he had not earned a copper all
day, and the gaily-dressed army of people swept selfishly past him,
thinking only of their Christmas dinners and warm homes.
The snowflakes fell from the leaden-coloured sky like great white
angels, to tell the earth that Christ would be born again on that night,
but Tom did not have any such ideas, as he was quite ignorant of
angels, and even of the birth of the child-Christ. He only looked upon
the snow as a cold and cruel thing, which made him shiver with pain,
and was a great trouble to brush away from his crossing.
 And overhead the mellow bells clashed out
their glad tidings in the bitterly chill air, while
below, in the warm, well-lighted churches, the
organ rolled out its hymns of praise, and the
worshippers said to one another, “Christ is
born again.”
But poor Tom!
Ah, how cold and hungry he was, standing in
the bright glare of the lamp, with his rags drawn closely round him for
protection against the falling snow. The throng of people grew
thinner and thinner, the gaily-decorated shops put up their shutters,
the lights died out in the painted windows of the churches, the bells
were silent, and only poor Tom remained in the deserted, lonely
streets, with the falling snowflakes changing him to a white statue.
He was thinking about going to his garret, when a gentleman,
wrapped in furs, passed along quickly, and just as he came near
Tom, dropped his purse, but, not perceiving his loss, walked on
rapidly through the driving snow. Tom’s first idea was to pick the
purse up and restore it to its owner, whom Tom knew very well by
sight, for he was a poet, who daily passed by Tom’s crossing. Then
Tom paused for a moment as he thought of all the beautiful things
the money in that purse would buy; while he hesitated, the poet
disappeared in the darkness of the night, so Tom was left alone with
the purse at his feet.
There it lay, a black object on the pure white
snow, and as Tom picked it up, he felt that it
was filled with money. Oh, how many things of
use to him could that money buy—bread and
meat and a cup of warm coffee—which would
do him good. Tom slipped it into his pocket, and thought he would
buy something to eat; but just at that moment he seemed to hear a
whisper in the air,—
 AS TOM PICKED IT UP HE FELT THAT IT WAS
FILLED WITH MONEY

“Thou shalt not steal.”

With a start of terror Tom looked around, thinking a policeman had
spoken, and would take him off to prison for stealing the purse, but
no policeman was in sight. He saw nothing but the whirling flakes
and his ragged shadow cast blackly on the white snow by the light of
the lamp. It could not have been the shadow speaking, as Tom
thought, for he knew that shadows never speak; but, ah! he did not
know the many wonderful things there are in this wonderful world of
ours.
Whoever had made the remark touched Tom’s heart, for he
remembered how his poor mother had blessed him when she died,
and told him to be an honest boy. It certainly would not be honest to
steal money out of the purse, but Tom was so cold and hungry that
he half thought he would do so. He took out the purse again and
looked at its contents—four shining sovereigns and some silver.
Then he put it back in his pocket, and trudged home with his broom
under his arm.
Home!—ah, what a dreary, cheerless home it was!—nothing but a
garret on the top of an old house—a bare garret, with no table or
chairs, but only the sacks upon which Tom slept at night.
He closed the door, and then lighted a little bit of candle he had
picked up in the streets with one of the matches from a box given
him by a ragged match-seller.
Tom placed the candle on the floor, and, kneeling down, opened the
purse to look at the money once more. Oh, how tempted he was to
take one of those shillings and buy some food and wood—it would
be a merry Christmas for him then! Other people were enjoying their
Christmas, and why should he not do the same? The great poet who
had dropped the purse had plenty of money, and would never miss
this small sum; so Tom, desperate with hunger, took a shilling, and,
hiding the purse under his bed, was about to blow out the candle
before creeping down-stairs to buy some food, when he heard a soft
voice whisper,—
“Don’t go, Tom.”
He turned round, and there was the shadow
cast by the reflection of the candle-light on the
wall. It was a very black shadow, much
blacker than Tom had ever seen before, and
as he looked it grew blacker and blacker on
the wall, then seemed to grow out of it until it
left the wall altogether, and stood by itself in
the centre of the floor, a waving, black shadow
of a ragged boy. Curiously enough, however,
Tom could not see its face, but only the outline
of its whole figure, yet it stood there shaking with every flicker of the
candle, and Tom could feel that its eyes were looking right into him.
“Don’t go, Tom,” said the shadow, in a voice so like his own that he
started. “If you go, you will be lost for ever.”
“Lost?” said Tom, with a laugh; “why, I couldn’t lose myself. I know
every street in the city.”
“I don’t mean really lost,” replied the shadow; “but it will be your first
step on the downward path.”
“Who are you?” asked Tom, rather afraid of the shadow, but keeping
a bold front.
“I am your shadow,” it replied, sighing. “I follow you wherever you go,
but only appear when there is light about you. If you had not lighted
that candle I would not have appeared, nor could I have spoken.”
“Was it you who spoke at the lamp-post?” said Tom doubtfully.
“Yes, it was I,” answered the shadow. “I wanted to save you then, as
I do now, from committing a crime. Sit down, Tom, and let us talk.”
Tom sat down, and the shadow sat down also. Then for the first time
he caught a passing glimpse of its face, just like his own, only the
eyes were sad—oh, so sad and mournful!
“Thou shalt not steal,” said the shadow solemnly.
“I don’t want to steal,” replied Tom sulkily; “but I’m cold and hungry.
This shilling would buy me fire and food. I don’t call that stealing.”
“Yes, but it is stealing,” answered the shadow, wringing its hands;
“and you know it is. If you steal you will be put in prison, and then I
shall have to go also. Think of that, Tom, think of that.”
Tom did not say a word, but sat on the floor looking at the bright
shilling in his hand which could procure him so many comforts. The
shadow saw how eager he was to take the shilling, and, with a sigh,
began to talk again.
“Think of your mother, Tom,” it said softly. “She was the wife of a
gentleman—your father; but he lost all his money, and when he died
she could get no one to help her. Do you remember how she died
herself in this very place, and how she implored you with her last
breath to be an honest boy?”
“Yes, I remember,” said Tom huskily; “but she did not know how cold
and hungry I would be.”
“Yes she did—she did,” urged the shadow. “She also had felt cold
and hunger, but she never complained. She never stole, and now
she has her reward, because she is a bright angel.”
“I don’t know what an angel is,” said Tom crossly; “but if she’s all
right, why doesn’t she help me?”
“She does help you, Tom,” said the shadow; “and it was because she
saw you were tempted to steal to-night that she asked me to help
you. She cannot speak as I do, because she is not a shadow.”
“Well, help me if you’re able,” said Tom defiantly; “but I don’t believe
you can.”
The candle on the floor had burnt very low, and as Tom said the last
words his shadow bent nearer and nearer, until he again saw those
mournful eyes, which sent a shiver through his whole body. It
stretched out its arms, and Tom felt them close round him like soft,
clinging mist; the candle flared up for a moment, and then went out,
leaving Tom in darkness altogether. But he did not feel a bit afraid,
for the soft arms of the shadow were round him, and he felt that it
was carrying him through the air.
They journeyed for miles and miles, but Tom knew not which
direction they were taking until a soft light seemed to spread all
around, and Tom felt that he was in the midst of a large crowd,
although he saw no one near him. Then he felt his bare feet touch
some soft, cloudy ground, that felt like a sponge; the shadowy arms
unclasped themselves, and he heard a voice, soft as the whispering
of winds in summer, sigh,—
“This is the Kingdom of Shadows.”
Then Tom’s eyes became accustomed to the subdued twilight, and
he saw on every side a number of shadows hurrying hither and
thither. He seemed to be in the centre of a wide plain, over which
hung a pale white mist, through which glimmered the soft light. The
shadows were all gliding about this plain; some thin, some fat, some
tall, others short; they all appeared to have business to do, and each
appeared to be intent only on his own concerns. Tom’s own shadow
kept close to him, and whispered constantly in his ear of strange
doings.
“These are the shadows of the past and of the future,” it sighed; “all
the shadows of human beings and their doings are here; see, there
is a funeral.”
And a funeral it was which came gliding over the smooth, white plain;
the great black hearse, the dark horses with nodding plumes, and
then a long train of mourners; all this came out of the mist at one
end, glided slowly over the plain, and vanished in the veil of mist at
the other. Then a bridal procession appeared; afterwards a great
army, clashing cymbals and blowing trumpets from whence no sound
of music proceeded; then the coronation triumph of a king, and later
on a confused multitude of men, women, and children, all hurrying
onward with eager rapidity. But they all came out of the mist and
went into the mist, only appearing on the white plain for a few
minutes, like the shadows of a magic lantern.
“The stage of the world,” whispered Tom’s shadow. “Birth, death, and
marriage, triumphs and festivities, joys and sorrows, all pass from
mist to mist, and none know whence they come or whither they go.”
“But what has this got to do with me?” asked Tom, who was feeling
rather bewildered.
“You are a man,” said his shadow reproachfully, “and must take an
interest in all that men do; but come, and I will show you what will
happen if you steal the purse.”
They glided over the plain towards the distant curtain of mist, but
how they travelled over the immense distance so rapidly Tom did not
know, for in a moment it seemed to him that he had come many
miles, and found himself suddenly before a grey, misty veil, with his
own shadow beside him, and many other shadows around.
As he stood there, a whisper like the murmur of the sea on a pebbly
beach sounded in his ears, and he seemed to guess, rather than
hear, what the shadows said.
“Now he will see—now he will see—he must choose the good or the
bad. Which will he choose?—which will he choose?”
Then the grey veil stirred, as if shaken by a gentle wind, and,
blowing aside, disclosed what seemed to Tom to be a great sheet of
ice of dazzling whiteness set up on end. As he looked, however,
shadows began to appear on the milky surface which acted a kind of
play and then vanished, and in the play he was always the central
figure.
First he saw himself pick up the purse in the snowy street; then hide
it in his bed. He saw his ragged shadow glide down-stairs from the
garret to buy food; the shopman looking at him, then at the shilling;
then a policeman arresting him and finding the purse hidden in the
bed. Afterwards he saw himself in prison; then released, and
prowling about the streets. Years seemed to pass as he looked, and
his shadow became taller and stouter, but always wearing a ragged
dress. After many years he seemed to see his shadow breaking into
a house—meet the owner of the house, and kill him. Afterwards the
shadow of himself stood in the dock; then crouched in prison; and,
last of all, he appeared standing under a black gallows with a rope
round his neck. At length all the shadows vanished, and the surface
of the ice mirror again became stainless, whilst a voice whispered in
his ear, “All this will happen if you steal the purse.”
Then the shadows again came on to the mirror and acted another
play; but this time it was much more pleasant.
Tom saw his shadow representative take the purse back to the poet
who had lost it. Then he saw himself in a school, learning all kinds of
wonderful things; and the years rolled by, as they had done in the
other play, unfolding the shadows of a beautiful life. He saw himself
become a great and famous poet, who wrote beautiful books to
make people wise and good. Then he saw himself in church, with a
woman’s shadow by his
side, and he knew, in some mysterious way, that it was the daughter
of the poet who had lost the purse. And as the happy years rolled on
he saw himself rich and honourable, and the end of all was a
magnificent funeral, taking his body to be buried in the great church
wherein many famous men were laid. Then the shadows vanished,
and the mirror became pure again, while over it the grey mists fell
like a soft veil, and once more the voice of his shadow said,—
“All this will happen if you remain honest.”
Then the crowd of shadows around Tom looked at him with their
mournful eyes, and a whispering question ran through the fantastic
throng,—
“Which will he choose?—which will he choose?”
“I will choose the honest life,” cried Tom loudly. “Yes, I will give back
the purse to the poet.”
At this the shadows around seemed to rejoice, and he could see
beautiful faces smiling at him from amid the crowd. The shadow
multitude broke in a wild dance of joy, keeping time to some aerial
music which Tom could not hear; and his own shadow, with
happiness shining out of its mournful eyes, threw its arms round him
once more. A dark veil seemed to fall over him, and the great white
plain, the glimmering mists, and the restless shadows, vanished
together.
When Tom opened his eyes again, he found himself lying on the
floor of his garret, cold and hungry still, but with his heart filled with a
great joy, for the shilling was still clutched in his hand, and he knew
he had not stolen the money. He took the purse from under the
sacks, replaced the shilling, and then went out, in the bright sunshine
of the Christmas morning, to give back the lost purse to its owner.
Overhead the bells rang out merrily, as if they were rejoicing at Tom’s
victory over himself, and a beautiful lady, who was on her way to
church, gave Tom some money to get food. He went and bought a
loaf and a cup of coffee, then, thankful for his good fortune, he
trudged off to the poet’s house.
The great poet received him very kindly, and,
after thanking Tom for returning his purse,
asked him why he had done so instead of
keeping it? Whereupon Tom told the poet all
about the shadow, which interested the poet
very much. He also had been to Shadowland
and seen strange things, which he told to the world in wonderful
verse.
“This boy is a genius,” he said to his wife, “and I must help him.”
Then it all happened as the magic mirror had foretold, for Tom was
put to school by the kind poet, and became a very clever man. He
also wrote poems, which the world received with joy; and when he
became a famous man, the kind poet gave him his own daughter in
marriage, and the bells which had rang the birth of the child-Christ
when Tom was a poor ragged boy, now rang out joyously in honour
of his marriage.
“He has conquered,” they clashed out in the warm, balmy air; “he is
the victor, and now he will be happy.”
And he was happy, very very happy, and felt deeply thankful to the
shadow who had shown him the way to be happy. His own shadow
never left him, but it never spoke to him again, though when Tom felt
tempted to do wrong, he heard a whisper advising him to do right.
Some people said that this was the voice of conscience, but Tom
knew it was the voice of his dear shadow, who still watched over
him.
And one day he took his wife to the garret where he had lived when
a poor boy, and told her how he had been to Shadowland, and
learned that to be honest and noble was the only true way to
happiness. His wife laughed, and said Tom had been dreaming; but
Tom shook his head, and said that it was no dream, but a great truth.
Now, who do you think was right—Tom or his wife?
THE WATER-WITCH
 I.

FIRE AND WATER.

ONCE upon a time, long long years ago, there was a shepherd
called Duldy, who dwelt in the forests which clothed the base of the
great mountain of Kel. This mountain was in the centre of an
immense plain, watered by many rivers, and dotted over with many
cities, for the kingdom of Metella was a very rich place indeed, so
rich that the inhabitants looked upon gold in the same way as we
look upon tin or iron, as quite a common thing. The plain was very
fertile by reason of the great rivers which flowed through it like silver
threads, and all these rivers took their rise in the mountain of Kel, a
mighty snow-clad peak which shot up, white and shining, to the blue
sky from amidst the bright green of its encircling forests.
There were old stories handed down from father to son, which said
that the mountain was once a volcano, which, breathing nothing but
fire, sent great streams of red-hot lava down to the fertile plain, to
wither and blight all the beautiful gardens and rich corn-fields. But
the fires in the breast of the mountain had long since died out, and
for many centuries the black, rugged summit had been covered with
snow, while countless streams, caused by the melting of the glaciers,
fell down its rocky sides, and, flowing through the cool, green pine
forests, spread themselves over the thirsty plain, so that it bloomed
like a beautiful garden.
Duldy lived in these scented pine forests, and
was supposed to be the son of an old couple
called Dull and Day, from whence by joining
both names he got his own Duldy; but he was
really a lost child whom old Father Dull had
found, seventeen years before, on the banks
of the Foam, one of the bright sparkling
streams which flowed from the snowy heights
above. Dull took the child home to his wife
Day, who was overcome with joy, for she