Experience Seamless Full Ebook Downloads for Every Genre at textbookfull.

com

Digital Image Processing A Signal Processing and

Algorithmic Approach 1st Edition D. Sundararajan
(Auth.)

https://textbookfull.com/product/digital-image-processing-a-
signal-processing-and-algorithmic-approach-1st-edition-d-
sundararajan-auth/

OR CLICK BUTTON

DOWNLOAD NOW

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

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

Digital Image Processing Practical Approach Borko Furht

https://textbookfull.com/product/digital-image-processing-practical-
approach-borko-furht/

textboxfull.com

From Algebraic Structures to Tensors Digital Signal and

Image Processing Matrices and Tensors in Signal Processing
Set 1st Edition Gérard Favier (Editor)
https://textbookfull.com/product/from-algebraic-structures-to-tensors-
digital-signal-and-image-processing-matrices-and-tensors-in-signal-
processing-set-1st-edition-gerard-favier-editor/
textboxfull.com

Digital Image Watermarking: Theoretical and Computational

Advances (Intelligent Signal Processing and Data Analysis)
1st Edition Borra
https://textbookfull.com/product/digital-image-watermarking-
theoretical-and-computational-advances-intelligent-signal-processing-
and-data-analysis-1st-edition-borra/
textboxfull.com

Conceptual Digital Signal Processing with MATLAB Keonwook

Kim

https://textbookfull.com/product/conceptual-digital-signal-processing-
with-matlab-keonwook-kim/

textboxfull.com
Digital Signal Processing-A Primer With MATLAB® 1st
Edition Samir I. Abood (Author)

https://textbookfull.com/product/digital-signal-processing-a-primer-
with-matlab-1st-edition-samir-i-abood-author/

textboxfull.com

Methods and techniques for fire detection : signal, image

and video processing perspectives 1st Edition Çetin

https://textbookfull.com/product/methods-and-techniques-for-fire-
detection-signal-image-and-video-processing-perspectives-1st-edition-
cetin/
textboxfull.com

Think DSP Digital Signal Processing in Python 1st Edition

Allen B. Downey

https://textbookfull.com/product/think-dsp-digital-signal-processing-
in-python-1st-edition-allen-b-downey/

textboxfull.com

Smartphone based real time digital signal processing Third

Edition Abhishek Sehgal

https://textbookfull.com/product/smartphone-based-real-time-digital-
signal-processing-third-edition-abhishek-sehgal/

textboxfull.com

Digital Image Enhancement Restoration and Compression

Digital Image Processing and Analysis for True Epub Scott
E Umbaugh
https://textbookfull.com/product/digital-image-enhancement-
restoration-and-compression-digital-image-processing-and-analysis-for-
true-epub-scott-e-umbaugh/
textboxfull.com
D. Sundararajan

Digital Image
Processing
A Signal Processing
and Algorithmic Approach
Digital Image Processing
D. Sundararajan

Digital Image Processing

A Signal Processing and Algorithmic
Approach

123
D. Sundararajan
Formerly at Concordia University
Montreal
Canada

Additional material to this book can be downloaded from http://extras.springer.com.

ISBN 978-981-10-6112-7 ISBN 978-981-10-6113-4 (eBook)
DOI 10.1007/978-981-10-6113-4
Library of Congress Control Number: 2017950001

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

Vision is one of our strongest senses. The amount of information conveyed through
pictures over the Internet and other media is enormous. Therefore, the field of
image processing is of great interest and rapidly growing. Availability of fast digital
computers and numerical algorithms accelerates this growth. In this book, the
basics of Digital Image Processing is presented, using a signal processing and
algorithmic approach. The image is a two-dimensional signal, and most processing
requires algorithms. Plenty of examples, figures, tables, programs, and physical
explanations make it easy for the reader to get a good grounding in the basics of the
subject, able to progress to higher levels, and solve practical problems.
The application of image processing is important in several areas of science and
engineering. Therefore, Digital Image Processing is a field of study for engineers
and computer science professionals. This book includes mathematical theory, basic
algorithms, and numerical examples. Thereby, engineers and professionals can
quickly develop algorithms and find solutions to image processing problems
of their interest using computers. In general, there is no formula for solving prac-
tical problems. Invariably, an algorithm has to be developed and used to find the
solution. While every solution is a combination of the basic principles, several
combinations are possible for solving the same problem. Out of these possibilities,
one has to come with the right solution. This requires some trial-and-error process.
A good understanding of the basic principles, knowledge of the characteristics
of the image data involved, and practical experience are likely to lead to an efficient
solution.
This book is intended to be a textbook for senior undergraduate- and
graduate-level Digital Image Processing courses in engineering and computer
science departments and a supplementary textbook for application courses such as
remote sensing, machine vision, and medical analysis. For image processing pro-
fessionals, this book will be useful for self-study. In addition, this book will be a
reference for anyone, student or professional, specializing in image processing. The
prerequisite for reading this book is a good knowledge of calculus, linear algebra,
one-dimensional digital signal processing, and programming at the undergraduate
level.

v
vi Preface

Programming is an important component in learning and practicing this subject.

A set of MATLAB® programs are available at the Web site of the book. While the
use of a software package is inevitable in most applications, it is better to use the
software in addition to self-developed programs. The effective use of a software
package or to develop own programs requires a good grounding in the basic
principles of the subject. Answers to selected exercises marked  are given at the
end of the book. A Solutions Manual and slides are available for instructors at the
Web site of the book.
I assume the responsibility for all the errors in this book and would very
much appreciate receiving readers’ suggestions and pointing out any errors
(email:[email protected]). I am grateful to my Editor and the rest of the
team at Springer for their help and encouragement in completing this project.
I thank my family for their support during this endeavor.

D. Sundararajan
About the Book

This book “Digital Image Processing—A Signal Processing and Algorithmic

Approach” deals with the fundamentals of Digital Image Processing, a topic of
great interest in science and engineering. Digital Image Processing is processing of
images using digital devices after they are converted to a 2-D matrix of numbers.
While the basic principles of the subject are those of signal processing, the appli-
cations require extensive use of algorithms. In order to meet these requirements, the
book presents the mathematical theory along with numerical examples with 4  4
and 8  8 subimages. The presentation of the mathematical aspects has been greatly
simplified with sufficient detail. Emphasis is given for physical explanation of the
mathematical concepts, which will result in deeper understanding and easier
comprehension of the subject. Further, the corresponding MATLAB codes are
given as supplementary material. The book is primarily intended as a textbook for
an introductory Digital Image Processing course at senior undergraduate and
graduate levels in engineering and computer science departments. Further, it can be
used as a reference by students and practitioners of Digital Image Processing.

vii
Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Image Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Digital Image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.1 Representation in the Spatial Domain . . . . . . . . . . . . . . 3
1.2.2 Representation in the Frequency Domain . . . . . . . . . . . 6
1.3 Quantization and Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.1 Quantization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3.2 Spatial Resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.3 Sampling and Aliasing . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.4 Image Reconstruction and the Moiré Effect . . . . . . . . . . 15
1.4 Applications of Digital Image Processing . . . . . . . . . . . . . . . . . . 16
1.5 The Organization of This Book . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2 Image Enhancement in the Spatial Domain . . . . . . . . . . . . . . . . . . . 23
2.1 Point Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.1.1 Image Complement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.1.2 Gamma Correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2 Histogram Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2.1 Contrast Stretching . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.2.2 Histogram Equalization . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.2.3 Histogram Specification . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.3 Thresholding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.4 Neighborhood Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.4.1 Linear Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.4.2 Median Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

ix
x Contents

3 Fourier Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.1 The 1-D Discrete Fourier Transform . . . . . . . . . . . . . . . . . . . . . . 66
3.2 The 2-D Discrete Fourier Transform . . . . . . . . . . . . . . . . . . . . . . 75
3.3 DFT Representation of Images . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.4 Computation of the 2-D DFT . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.5 Properties of the 2-D DFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.6 The 1-D Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
3.7 The 2-D Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4 Image Enhancement in the Frequency Domain. . . . . . . . . . . . . . . . . 109
4.1 1-D Linear Convolution Using the DFT . . . . . . . . . . . . . . . . . . . 110
4.2 2-D Linear Convolution Using the DFT . . . . . . . . . . . . . . . . . . . 111
4.3 Lowpass Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.3.1 The Averaging Lowpass Filter . . . . . . . . . . . . . . . . . . . 112
4.3.2 The Gaussian Lowpass Filter . . . . . . . . . . . . . . . . . . . . . 117
4.4 The Laplacian Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
4.4.1 Amplitude and Phase Distortions . . . . . . . . . . . . . . . . . . 124
4.5 Frequency-Domain Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.5.1 Ideal Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.5.2 The Butterworth Lowpass Filter . . . . . . . . . . . . . . . . . . 129
4.5.3 The Butterworth Highpass Filter . . . . . . . . . . . . . . . . . . 132
4.5.4 The Gaussian Lowpass Filter . . . . . . . . . . . . . . . . . . . . . 134
4.5.5 The Gaussian Highpass Filter . . . . . . . . . . . . . . . . . . . . 134
4.5.6 Bandpass and Bandreject Filtering. . . . . . . . . . . . . . . . . 134
4.6 Homomorphic Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
4.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
5 Image Restoration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
5.1 The Image Restoration Process . . . . . . . . . . . . . . . . . . . . . . . . . . 143
5.2 Inverse Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
5.3 Wiener Filter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
5.3.1 The 2-D Wiener Filter . . . . . . . . . . . . . . . . . . . . . . . . . . 150
5.4 Image Degradation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.5 Characterization of the Noise and Its Reduction . . . . . . . . . . . . . 154
5.5.1 Uniform Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
5.5.2 Gaussian Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
5.5.3 Periodic Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
5.5.4 Noise Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
5.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Contents xi

6 Geometric Transformations and Image Registration . . . . . . . . . . . . 163

6.1 Interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
6.1.1 Nearest-Neighbor Interpolation . . . . . . . . . . . . . . . . . . . 164
6.1.2 Bilinear Interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
6.2 Affine Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
6.2.1 Scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
6.2.2 Shear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6.2.3 Translation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
6.2.4 Rotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
6.3 Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
6.3.1 1-D Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
6.3.2 2-D Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
6.4 Image Registration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
7 Image Reconstruction from Projections . . . . . . . . . . . . . . . . . . . . . . . 189
7.1 The Normal Form of a Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
7.2 The Radon Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
7.2.1 Properties of the Radon Transform . . . . . . . . . . . . . . . . 196
7.2.2 The Discrete Approximation of the Radon
Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
7.2.3 The Fourier-Slice Theorem . . . . . . . . . . . . . . . . . . . . . . 202
7.2.4 Reconstruction with Filtered Back-projections . . . . . . . . 206
7.3 Hough Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
7.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
8 Morphological Image Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
8.1 Binary Morphological Operations . . . . . . . . . . . . . . . . . . . . . . . . 218
8.1.1 Dilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
8.1.2 Erosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
8.1.3 Opening and Closing . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
8.1.4 Hit-and-Miss Transformation . . . . . . . . . . . . . . . . . . . . . 229
8.1.5 Morphological Filtering . . . . . . . . . . . . . . . . . . . . . . . . . 231
8.2 Binary Morphological Algorithms . . . . . . . . . . . . . . . . . . . . . . . 233
8.2.1 Thinning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
8.2.2 Thickening. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
8.2.3 Noise Removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
8.2.4 Skeletons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
8.2.5 Fill. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
8.2.6 Boundary Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
8.2.7 Region Filling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
8.2.8 Extraction of Connected Components . . . . . . . . . . . . . . 244
xii Contents

8.2.9 Convex Hull . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244

8.2.10 Pruning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
8.3 Grayscale Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
8.3.1 Dilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
8.3.2 Erosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
8.3.3 Opening and Closing . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
8.3.4 Top-Hat and Bottom-Hat Transformations . . . . . . . . . . . 249
8.3.5 Morphological Gradient . . . . . . . . . . . . . . . . . . . . . . . . . 250
8.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
9 Edge Detection . . . . . . . . . . . . . . . . . . . . . . ................. . . . . 257
9.1 Edge Detection . . . . . . . . . . . . . . . . . ................. . . . . 257
9.1.1 Edge Detection by Compass Gradient Operators . . . . . . 264
9.2 Canny Edge Detection Algorithm . . . ................. . . . . 266
9.3 Laplacian of Gaussian. . . . . . . . . . . . ................. . . . . 273
9.4 Summary . . . . . . . . . . . . . . . . . . . . . ................. . . . . 278
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . ................. . . . . 278
10 Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
10.1 Edge-Based Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
10.1.1 Point Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
10.1.2 Line Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
10.2 Threshold-Based Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . 284
10.2.1 Thresholding by Otsu’s Method . . . . . . . . . . . . . . . . . . 286
10.3 Region-Based Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
10.3.1 Region Growing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
10.3.2 Region Splitting and Merging . . . . . . . . . . . . . . . . . . . . 293
10.4 Watershed Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
10.4.1 The Distance Transform . . . . . . . . . . . . . . . . . . . . . . . . 295
10.4.2 The Watershed Algorithm . . . . . . . . . . . . . . . . . . . . . . . 300
10.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
11 Object Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
11.1 Boundary Descriptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
11.1.1 Chain Codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
11.1.2 Signatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
11.1.3 Fourier Descriptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312
11.2 Regional Descriptors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
11.2.1 Geometrical Features . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
11.2.2 Moments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
11.2.3 Textural Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
Contents xiii

11.3 Principal Component Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 334

11.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
12 Object Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
12.1 The k-Nearest Neighbors Classifier . . . . . . . . . . . . . . . . . . . . . . . 345
12.2 The Minimum-Distance-to-Mean Classifier . . . . . . . . . . . . . . . . . 347
12.2.1 Decision-Theoretic Methods . . . . . . . . . . . . . . . . . . . . . 349
12.3 Decision Tree Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
12.4 Bayesian Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
12.5 k-Means Clustering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356
12.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
13 Image Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
13.1 Lossless Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
13.1.1 Huffman Coding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
13.1.2 Run-Length Encoding . . . . . . . . . . . . . . . . . . . . . . . . . . 368
13.1.3 Lossless Predictive Coding . . . . . . . . . . . . . . . . . . . . . . 369
13.1.4 Arithmetic Coding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
13.2 Transform-Domain Compression . . . . . . . . . . . . . . . . . . . . . . . . 382
13.2.1 The Discrete Wavelet Transform . . . . . . . . . . . . . . . . . . 383
13.2.2 Haar 2-D DWT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
13.2.3 Image Compression with Haar Filters . . . . . . . . . . . . . . 389
13.3 Image Compression with Biorthogonal Filters . . . . . . . . . . . . . . 391
13.3.1 CDF 9/7 Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
13.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403
14 Color Image Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407
14.1 Color Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
14.1.1 The RGB Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
14.1.2 The XYZ Color Model . . . . . . . . . . . . . . . . . . . . . . . . . 412
14.1.3 The CMY and CMYK Color Models . . . . . . . . . . . . . . 412
14.1.4 The HSI Color Model . . . . . . . . . . . . . . . . . . . . . . . . . . 414
14.1.5 The NTSC Color Model . . . . . . . . . . . . . . . . . . . . . . . . 419
14.1.6 The YCbCr Color Model. . . . . . . . . . . . . . . . . . . . . . . . 420
14.2 Pseudocoloring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
14.2.1 Intensity Slicing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
14.3 Color Image Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
14.3.1 Image Complement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424
14.3.2 Contrast Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . . 426
14.3.3 Lowpass Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427
14.3.4 Highpass Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428
14.3.5 Median Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
xiv Contents

14.3.6 Edge Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429

14.3.7 Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
14.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
Appendix A: Computation of the DFT . . . . . . . . . . . . . . . . . . . . . . . . . . . 439
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
Answers to Selected Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
About the Author

D. Sundararajan is a full-time author in signal processing and related areas. In

addition, he conducts workshops on image processing, MATLAB, and LATEX. He
was formerly associated with Concordia University, Montreal, Canada, and other
universities and colleges in India and Singapore. He holds a M.Tech. degree in
Electrical Engineering from Indian Institute of Technology, Chennai, India, and a
Ph.D. degree in Electrical Engineering from Concordia university, Montreal,
Canada. His specialization is in signal and image processing. He holds a US, a
Canadian, and a British Patent related to discrete Fourier transform algorithms. He
has written four books, the latest being “Discrete wavelet transform, a signal pro-
cessing approach” published by John Wiley (2015). He has published papers in
IEEE transactions and conferences. He has also worked in research laboratories in
India, Singapore, and Canada.

xv
Abbreviations

1-D One-dimensional
2-D Two-dimensional
3-D Three-dimensional
bpp Bits per pixel
DC Sinusoid with frequency zero, constant
DFT Discrete Fourier transform
DWT Discrete wavelet transform
FIR Finite impulse response
FT Fourier transform
IDFT Inverse discrete Fourier transform
IDWT Inverse discrete wavelet transform
IFT Inverse Fourier transform
LoG Laplacian of Gaussian
LSB Least significant bit
MSB Most significant bit
PCA Principal component analysis
SNR Signal-to-noise ratio

xvii
Chapter 1
Introduction

Abstract The image of a scene or object is inherently a continuous two-dimensional

signal. Due to the advantages of digital systems, this type of image has to be con-
verted into a discrete signal. This change in form requires sampling and quantization.
The characteristics of a digital image and its spatial- and frequency-domain repre-
sentations are introduced. The sampling and quantization operations are described.

Most of the information received by humans is visual. A picture is a 2-D visual

representation of a 3-D scene. A picture is worth a thousand words. That is, a certain
amount of information can be quickly and effectively conveyed by a picture. It is
obvious from the popularity of the film medium, Internet, and digital cameras. Digital
image processing is the processing of images using digital computers and is used in
many applications of science and engineering. It is implied that natural images are
converted to digital form prior to processing.
While an image is a 2-D signal, a considerable amount of its processing is carried
out in one dimension (row by row and column by column). Therefore, we start with 1-
D signals. An example of a one-dimensional (1-D) signal is x(t) = sin(t). x(t) is the
amplitude of the signal at t, the independent variable. Variable t is usually associated
with continuous time. As most of the practical signals are of continuous type and
digital signal processing is advantageous, the signal is sampled and quantized. A
1-D discrete signal is usually specified as x(n), where the independent variable n
is an integer. The sampling interval Ts is usually suppressed. A discrete image is
a two-dimensional (2-D) signal, x(m, n), where m and n are the two independent
variables. The amplitude of the image x(m, n) at each point is called the pixel value.
Pixel stands for picture element. The three major goals of digital image processing
are: (i) to improve the quality of the image for human perception, (ii) to improve the
quality and represent the image suitable for automatic machine perception, and (iii)
to compress the image so that the storage and transmission requirements are reduced.
The requirements for human and machine perceptions are, in general, different. These
tasks are carried out by computers after the picture is represented in a numeric form.
The use of digital cameras, which directly produce digital images, is in prevalent
use. Scanners are available to digitize analog photographs. With some exceptions,
the processing of an image, which is a 2-D signal, is a straightforward extension of

© Springer Nature Singapore Pte Ltd. 2017 1

D. Sundararajan, Digital Image Processing, DOI 10.1007/978-981-10-6113-4_1
2 1 Introduction

Table 1.1 Electromagnetic spectrum

Cosmic Gamma X-rays Ultra Visible Infra-red Microwaves TV Radio
rays rays violet spectrum

that of 1-D signals. For example, with a good knowledge of important operations
such as sampling, convolution, and Fourier analysis of 1-D signals, one can easily
adapt to their extension for 2-D signals.
An image is some form of a picture giving a visual representation of a scene or
an object for human or machine perception. Light is an electromagnetic radiation
that can produce visual sensation. Photon is a quantum of electromagnetic radiation.
Photons travel at the speed of light. The wavelength of the electromagnetic spectrum
varies from λ = 10−12 to 103 m. Components of the electromagnetic spectrum are
shown in Table 1.1. Frequency f in Hz and wavelength λ in meters are related by
the expression
(2.998)108
f =
λ
High-frequency photons carry more energy than the low-frequency photons. That
small part of the spectrum from λ = (0.43)10−6 to (0.79)10−6 m, which is visible
for human beings, is called the visible spectrum.
The invisible part of the spectrum is also of interest in image processing, since it
can be sensed by machines (e.g., X-ray is important in the medical field). As in the case
of most naturally occurring signals, an image is also a continuous signal inherently.
This signal has to be sampled and quantized to make it a digital image. Except
that there are two frequency components in two directions to be considered, the
sampling is governed by the 1-D sampling theorem. Both sampling and quantization
are constrained by the two contradicting criteria, accuracy and processing time.
Each point in an image corresponds to a small part of the scene making the
image. The brightness of the light received by an observer from a scene varies as
the reflectivity of the objects composing the scene and the illumination vary. This
type, which is most common, is called a reflection image. Another type, called the
emission image, is obtained from self-luminous objects such as stars or lights. A third
type, called the absorption image, is the result of radiation passing through objects.
The variation of the attenuation of the intensity of radiation (such as X-ray) recorded
by a film is the image. While camera produces most of the images, images are also
formed by other sensors such as infrared and ultrasonic. Irrespective of the source,
the processing of images involves the same basic principles.

1.1 Image Acquisition

The visual information is a function of two independent variables. It is a 2-D signal.

Nowadays, digital cameras produce digital images. These cameras use some type of
array of photosensitive devices to produce electrical signals proportional to the scene
1.1 Image Acquisition 3

brightness over small patches of a scene. The incident light on these devices create
charge carriers (holes and electrons), and a voltage applied across the device causes
the conduction of current. The potential difference across a resistor in the path of this
current is proportional to the average intensity of the light received by the device. The
resulting voltages of the array represent the scene being captured as an image. The
set of analog signals is converted to a digital image by an interface, called the frame
grabber. This interface is a constituent part of digital cameras, and the digital image
is delivered in a standard format through an interface to the computer. Of course, it
is understood that the sampling and quantization resolutions are set as required, at
the time of taking the picture. Invariably, the digital image requires some processing
either to enhance it with respect to some criteria and/or to extract useful information
for various applications. That is digital image processing. In this chapter, we study
the form and characteristics of the digital image.

1.2 Digital Image

While a scene is typically three dimensional, it is represented in two dimensions in

the image. In digital image processing, an image is represented as a 2-D matrix of
numbers. A M × N image with M rows and N columns is given by

n→
m⎡ x(0, 0) x(0, 1) x(0, 2) . . .
x(0, N − 1)
⎤
↓⎢ x(1, 0) x(1, 1) x(1, N − 1) ⎥
x(1, 2) . . .
x(m, n) = ⎢ ⎥
⎢ .. ⎥
⎣ . ⎦
x(M − 1, 0) x(M − 1, 1) x(M − 1, 2) . . . x(M − 1, N − 1)

With reference to the image, the pixel (picture element) located at (m, n) is with
value x(m, n). The image coordinates are m and n, and x(m, n) is proportional to
the brightness of the scene about that point. This domain of representation is called
the spatial domain, similar to the representation of a 1-D discrete signal in the time
domain. While the top-left corner is the origin in most cases, sometimes we also use
the bottom-left corner as the origin.

1.2.1 Representation in the Spatial Domain

An image is usually represented in the spatial domain by three forms. A 1-D signal,
such as the sine waveform y(t) = sin(t), is a curve, and we are familiar with
its representation in a figure with t represented by the x-axis and y(t) = sin(t)
represented by the y-axis. The independent variable is t and y(t) is the dependent
variable because the values of y(t) depend on the values of t. While a 1-D signal is a
4 1 Introduction

(a) (b)
255
50

100

x(m,n)
m

150

0
200
200
200
250 100
100
50 100 150 200 250
m 0 0 n
n

Fig. 1.1 a A 256 × 256 image with 256 gray levels; b its amplitude profile

(a) (b)

50 50

100 100
m

150 150

200 200

250 250
50 100 150 200 250 50 100 150 200 250
n n

Fig. 1.2 a A 256 × 256 image with its intensity values increasing, for each row, from 0 to 255; b
A 256 × 256 synthetic image with 256 gray levels

curve, a 2-D signal is a surface. Therefore, an image x(m, n) can be represented as a

surface with the m- and n-axes fixing the two coordinates and a third axis fixing its
amplitude. Figure 1.1a shows a 256 × 256 image with 256 gray levels and (b) shows
its amplitude profile. While the amplitude profile is mostly used to represent 1-D
signals, images are mostly represented using the intensity of its pixels.
Figure 1.1a is the representation of an image by the intensity (gray level) values
of its pixels. In a monochromatic or gray-level image, typically, a byte of storage is
used to represent the pixel value. With 8 bits, the pixel values are integers from 0 to
255. Figure 1.2a shows a 256 × 256 image in which the intensity values, for each
row, are increasing from 0 to 255. The value of all the pixels in the first row is 0,
those of the second row is 1, and so on. The value of all the pixels in the bottom
1.2 Digital Image 5

Table 1.2 Pixel values of a 8 × 8 subimage

173 185 189 186 199 195 195 192
177 187 189 192 197 195 189 177
188 190 196 197 199 193 171 124
191 192 197 198 192 158 111 110
196 199 99 189 149 108 110 113
202 200 182 130 100 98 108 114
204 178 117 85 100 96 104 108
173 100 85 87 95 98 96 100

row is 255. Starting from black in the top, the image gradually becomes white at the
bottom. Typically, zero is black and the maximum value is white. The value of all the
white pixels is set to 255, and the value of the black ones is set to zero. The values
between zero and the maximum value are shades of gray (a color between white and
black).
A simple image is shown in Fig. 1.2b, which is composed of three squares, with
various gray levels, in a black background. This is a synthetic image. This type of
images is useful for algorithm design, development, debugging, and verification,
since their values and the output of the algorithms are easily predictable. The image
has 256 rows of pixels, and each row is made up of 256 pixels with the gray level
varying from 0 to 255. The gray-level values of the three squares, from top to bottom,
are 84, 168, and 255, respectively.
Another representation of an image is by the numerical values of its intensity, as
shown in Table 1.2. While it is impossible to represent a large image in this form,
it is, in addition to synthetic images, extremely useful in algorithm development,
debugging, and verification (which is a major task in image processing applications)
with subimages typically of sizes 4 × 4 and 8 × 8.
In a color image, each pixel is vector-valued. Typically, a color pixel requires
24 bits of storage. A color image is a combination of images with basis colors. For
example, a color image is composed of its red, green, and blue components. If each
component is represented with 8 bits, then a color pixel requires 24 bits. While most
of the natural images are color images, the processing of gray-level images is given
importance because its processing can be easily extended to color images in most
cases and gray-level images contain essential information of the image. In a binary
image, a pixel value is stored in a bit, 0 or 1. Typical binary images contain text,
architectural plans, and fingerprints.
When operations, such as transforms, are carried out on images, the resulting
images may have widely varying amplitude range and precision. In such cases,
quantization is required. More often, images are square and typical sizes vary from
256 × 256 to 4096 × 4096. The numbers are usually a power of 2. Image processing
operations are easier with these numbers. For example, in order to reduce the size of
6 1 Introduction

an image to one-half, we simply discard alternate pixels. The all-important Fourier

analysis is carried out, in practice, with these numbers.
Some examples of the requirement of storage for images are:
(i) 512 × 512 binary image,

512 × 512 × 1 = 262144 bits = 32768 bytes

(ii) 512 × 512 8-bit gray-level image,

512 × 512 × 1 = 262144 bytes

(iii) 512 × 512 color image, with a byte of storage for each of the three color
components of a pixel,

512 × 512 × 3 = 786432 bytes

While the picture quality improves with increasing the size, the execution time of
algorithms also increases at a fast rate. Therefore, the minimum size that satisfies
the application requirements should be selected. The selection of fast algorithms is
also equally critical. Even with the modern computers, processing of images could
be slow depending on the size of the image and the complexity of the algorithm
being executed. Therefore, the minimum size, the simplest type (binary, gray-level,
or color image), and an appropriate algorithm must be carefully chosen for efficient
and economical image processing for any application.

1.2.2 Representation in the Frequency Domain

One of the major tasks in image processing is to find suitable representations of

images in other domains, in addition to the spatial domain, so that the processing
becomes easier and efficient, as is the case in 1-D signal processing. The suitable
representation invariably requires taking the transform of the image. Transforms
approximate practical images, which usually have arbitrary amplitude profiles, as a
weighted sum of a finite set of well-defined basis functions with adequate accuracy.
There are many transforms used in image processing, and each one has a different
set of basis functions and is suitable for some tasks. The most important of all the
transforms is the Fourier transform. Sinusoidal curves are the Fourier basis func-
tions for 1-D signals, and sinusoidal surfaces, such as that shown in Fig. 1.3, are the
Fourier basis functions for 2-D signals (images). In a transformed form, important
characteristics of the images, such as their frequency content, can be estimated. The
interpretation of operations, such as filtering of images, becomes easier. Further, the
computational complexity of operations and storage requirements are also reduced
in most cases.
1.3 Quantization and Sampling 7

Fig. 1.3 A 64 × 64
sinusoidal surface, which is a
typical basis function in the
2-D Fourier transform 200
representation of images

x(m,n)
0

−200

60
40 60
40
20
20
m 0 0 n

1.3 Quantization and Sampling

Sampling is required due to limited spatial and temporal resolutions (number of

pixels) of a digital image. Quantization is required due to limited intensity resolution
(wordlength). A pixel value, typically, is the integral of the image intensity over a
finite area. As most practical signals are continuous functions of continuous variables,
both sampling and quantization are required to get a digital signal so that they can
be processed by a digital computer. Sampling is converting a continuous function
into a discrete one. The values of a sampled function are known only at the discrete
values of its independent variable. Quantization is converting a continuous variable
into a discrete one. The values of a quantized variable are fixed at discrete intervals.
Consider one period of the continuous sinusoidal signal

2π π
x(t) = cos t+
16 6

shown in Fig. 1.4. The signal is sampled with a sampling interval of 1 s. Therefore,
starting with t = 0, we get 16 samples

x(n) ={0.8660, 0.6088, 0.2588, −0.1305, −0.5000, −0.7934, −0.9659, −0.9914,

− 0.8660, −0.6088, −0.2588, 0.1305, 0.5000, 0.7934, 0.9659, 0.9914}

These samples are further quantized with a quantization step of 0.2. That is, each
sample value is restricted to one of the finite set of values

{1, 0.8, 0.6, 0.4, 0.2, 0, −0.2, −0.4, −0.6, −0.8, −1}
8 1 Introduction

Fig. 1.4 Sampling and 1

quantizing a 1-D signal
0.8
0.6 quantized
actual
0.4
0.2

x(t)
0
−0.2
−0.4
−0.6
−0.8
−1
0 4 8 12
t

Each sample is assigned to the nearest allowed value. The samples of the sampled
and quantized signal are

xq (n) ={0.8, 0.6, 0.2, −0.2, −0.6, −0.8, −1.0, −1.0, −0.8, −0.6,
− 0.2, 0.2, 0.6, 0.8, 1.0, 1.0}

shown by dots in Fig. 1.4. The actual sample values are shown by crosses. Maximum
error is one-half of the quantization step. Both sampling and quantization operations
introduce errors in the representation of a signal. According to the sampling theorem,
the sampling frequency has to be more than twice that of the highest frequency content
of the signal. The quantization step should be selected so that the quantization noise
is within acceptable limit.

1.3.1 Quantization

Quantization is the process of mapping the amplitude of a continuous variable into a

set of finite discrete values. For a digital representation, the pixel values of an image
have to be quantized to some finite levels so that the image can be stored using a
finite number of bits. Typically, 8 bits are used to represent a pixel value. Figure 1.5
shows the effect of quantization of the pixel values, using 8, 7, 6, 5, 4, 3, 2, and 1 bits.
Reducing the number of bits reduces the number of gray levels and, in turn, reduces
the contrast of the image. The deterioration in quality is not noticeable upto 6 bits of
representation. From 5 bits onward, grayscale contouring effect is noticeable. False
edges appear when the gradually changing pixel values in a region of the image are
replaced by a single value. Due to the lower quantization levels, edges are created
between adjacent regions. As the use of 6 or 7 bits does not save much and the
8-bit (byte) wordlength is in popular use in the computer architectures, the 8-bit
representation is most often used.
1.3 Quantization and Sampling 9

Fig. 1.5 Representations of 8 7

an image using 8, 7, 6, 5, 4,
3, 2, and 1 bits

6 5

4 3

2 1
10 1 Introduction

Fig. 1.6 Bit-plane bit−plane 7:MSB bit−plane 6

representations of an image

bit−plane 5 bit−plane 4

bit−plane 3 bit−plane 2

bit−plane 1 bit−plane 0:LSB

1.3 Quantization and Sampling 11

The relative influence of the various bits in the formation of the image is shown in
Fig. 1.6. A gray-level image can be decomposed into a set of binary images, which
is useful in applications such as compression. The last image corresponds to the
least significant bit. It looks like an image generated by a set of random numbers,
and it is difficult to relate it to the original image. Higher-order bits carry more
information. As expected, the most significant bit carries most information and the
corresponding image (the first) resembles like its original. The bit-plane images can
be isolated from the grayscale images by repeatedly dividing the image matrix by
successive powers of 2 and taking the remainder of dividing the truncated quotient by
2. For example, let x = {0, 1, 2, 3, 4, 5}. Dividing x by 2 and taking the remainders,
we get x0 = {0, 1, 0, 1, 0, 1}. Dividing x by 2 and taking the truncated quotients,
we get x2q = {0, 0, 1, 1, 2, 2}. Dividing x2q by 2 and taking the remainders, we
get x1 = {0, 0, 1, 1, 0, 0}. Dividing x by 4 and taking the truncated quotients, we
get x4q = {0, 0, 0, 0, 1, 1}. Dividing x4q by 2 and taking the remainders, we get
x2 = {0, 0, 0, 0, 1, 1}. Note that 20 x0 + 21 x1 + 22 x2 = x. The 4 × 4 4-bit image
x(m, n) and its bit-plane components from MSB to LSB are
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
8 1 7 3 100 0 0 0 1 0 001 1 011 1
⎢1 11 15 12 ⎥ ⎢0 1 1 1⎥ ⎢0 0 1 1⎥ ⎢0 1 1 0⎥ ⎢1 1 1 0⎥
⎢ ⎥=2 ⎢
3 ⎥+2 ⎢
2 ⎥+2⎢ ⎥+⎢ ⎥
⎣0 11 7 13 ⎦ ⎣0 1 0 1⎦ ⎣0 0 1 1⎦ ⎣0 1 1 0⎦ ⎣0 1 1 1⎦
2 10 9 6 011 0 0 0 0 1 110 1 001 0

Quantization levels with equal intervals is called linear quantization. In nonlinear

quantization, the range of the frequently occurring pixels is quantized using more bits
and vice versa. The average error due to quantization is reduced without increasing
the number of bits. This type of quantization is often used in image compression.
It should be noted that, while sampling and quantization are necessary to get the
advantages of digital image processing, the image is corrupted to some extent due
to the quantization noise and the aliasing effect. It should be ensured that image
quality is within acceptable limits by proper selection of the sampling interval and
the quantization levels. In general, a rapidly varying scene requires a higher sampling
rate and fewer quantization levels and vice versa. A 256 × 256 image with 64 gray
levels is typically the minimum for most practical purposes.

1.3.2 Spatial Resolution

An image represents an object of a certain area. The spatial resolution is the physical
area of the object represented by a pixel. The resolution varies from nanometers in
microscopic images to kilometers in satellite images. The number of independent
pixel values per unit distance (pixel density) indicates the spatial resolution. A higher
number of pixels improves the ability to see finer details of an object in the image. For
example, the resolution of a digital image of size 512 × 512 formed from an analog
image of size 32×32 cm is 512/32 = 16 pixels per centimeter. Figure 1.7a–d shows,
12 1 Introduction

(a) (b)

64 32

128 64

192 96

64 128 192 32 64 96

(c) (d)

16 8

32 16

48 24

16 32 48 8 16 24

Fig. 1.7 Effects of reducing the spatial resolution. a Resolution 256 × 256; b resolution 128 × 128;
c resolution 64 × 64; d resolution 32 × 32

respectively, an image with resolutions 256 × 256, 128 × 128, 64 × 64, and 32 × 32.
Reducing the spatial resolution results in blockiness of the image. The blockiness
is just noticeable in the image in (b) and clearly seen in the image in (c), while the
image in (d) becomes unrecognizable.

1.3.3 Sampling and Aliasing

When sampling a signal, the sampling frequency must be greater than twice that of
its highest frequency component in order to reconstruct the signal perfectly from
its samples. In the case of an image, there are two frequency components (horizon-
tal and vertical) to be considered. Aliasing effect is the impersonation of a higher
1.3 Quantization and Sampling 13

frequency sinusoid as a lower frequency sinusoid due to insufficient number of sam-

ples. An arbitrary sinusoid with frequency f Hz requires more than 2 f samples for
its unambiguous representation by its samples. Aliasing can be eliminated by suitable
lowpass filtering of the image and then sampling so that the bandwidth is less than
half of that of the sampling frequency. The price that is paid for eliminating aliasing
is the blurring of the image, since high-frequency components, which provide the
details, are removed in lowpass filtering.
The aliasing effect is characterized by the following formulas.
 
2π 2π N
x(n) = cos (k + l N )n + φ = cos kn + φ , k = 0, 1, . . . , −1
N N 2
 
2π 2π N
x(n) = cos (l N − k)n + φ = cos kn − φ , k = 1, 2, . . . , −1
N N 2

where the number of samples N and index l are positive integers. With N even, oscil-
lations increase only upto k = N2 , decrease afterward, and cease at k = N , and this
pattern repeats indefinitely. With frequency indices higher than N2 , frequency fold-
ing occurs. Therefore, sinusoids with frequency index upto N2 can only be uniquely
identified with N samples. Frequency with index N2 is called the folding frequency.
The implication is that, with the number of samples fixed, only a limited number of
sinusoidal components can be distinctly identified. For example, with 256 samples,
the uniquely identifiable frequency components are

2π
x(n) = cos kn + φ , k = 0, 1, . . . , 127
256

Figure 1.8a shows a 32 × 32 sinusoidal surface


2π 2π π
x(m, n) = cos 2m + 1n +
32 32 2

with frequencies 2/32 and 1/32 cycles per sample along the m and n axes, respectively.
The bottom peak of the sinusoidal surface is black, and the top peak is white. It is
clear that the surface makes 2 cycles along the m axis and one along the n axis.
Consider a 32 × 32 sinusoidal surface

2π 2π π
x(m, n) = cos 30m + 31n −
32 32 2

with frequencies 30/32 and 31/32 cycles per sample along the m and n axes, respec-
tively. Frequency with index N2 = 32 2
= 16 is called the folding frequency. The
apparent frequencies are (32 − 30)/32 = 2/32 and (32 − 31)/32 = 1/32 cycles
per sample, respectively. This sinusoidal surface also produces oscillations with the
same frequency as in Fig. 1.8a.
14 1 Introduction

(a) 0 (b) 0
4
4
8
8
12
12
16
m

m
16
20
20
24
24
28
28

0 4 8 12 16 20 24 28 0 4 8 12 16 20 24 28
n n
π
Fig. 1.8 Aliasing effect. a x(m, n) = cos( 2π
32 2m + 32 1n + 2 ) = cos( 32 30m +
2π 2π 2π
32 31n − π2 ); b
x(m, n) = cos( 32 4m + 32 7n + π) = cos( 32 28m + 32 25n − π)
2π 2π 2π 2π


2π 2π π
x(m, n) = cos 30m + 31n −
32 32 2
 
2π 2π π 2π 2π π
= cos (32 − 2)m + (32 − 1)n − = cos 2m + 1n +
32 32 2 32 32 2

Figure 1.8b shows a 32 × 32 sinusoidal surface


2π 2π
x(m, n) = cos 4m + 7n + π
32 32

with frequencies 4/32 and 7/32 cycles per sample along the m and n axes, respectively.
It is clear that the surface makes 4 cycles along the m axis and 7 along the n axis.
Consider the sinusoidal surface

2π 2π
x(m, n) = cos 28m + 25n − π
32 32
 
2π 2π 2π 2π
= cos (32 − 4)m + (32 − 7)n − π = cos 4m + 7n + π
32 32 32 32

This sinusoidal surface also produces oscillations with the same frequency as in
Fig. 1.8b. If we double the number of samples, then aliasing is avoided in these
cases.
To fix the sampling frequency for a class of real-valued images, find the Fourier
spectra of typical images using the 2-D DFT for increasing sampling frequencies.
The appropriate sampling frequency in each of the two directions is that which
yields negligible spectral magnitude values in the vicinity of one-half of the sampling
frequency.
Exploring the Variety of Random
Documents with Different Content
The Project Gutenberg eBook of The judging of
the priestess
This ebook is for the use of anyone anywhere in the United States
and most other parts of the world at no cost and with almost no
restrictions whatsoever. You may copy it, give it away or re-use it
under the terms of the Project Gutenberg License included with this
ebook or online at www.gutenberg.org. If you are not located in the
United States, you will have to check the laws of the country where
you are located before using this eBook.

Title: The judging of the priestess

Author: Nelson S. Bond

Illustrator: Julian S. Krupa

Release date: July 12, 2024 [eBook #74025]

Language: English

Original publication: Chicago, IL: Ziff-Davis Publishing Company,

1940

Credits: Greg Weeks, Mary Meehan and the Online Distributed

Proofreading Team at http://www.pgdp.net

*** START OF THE PROJECT GUTENBERG EBOOK THE JUDGING OF

THE PRIESTESS ***
The Judging of the Priestess

By NELSON S. BOND

Out of future Mexico the Japcans came,

invading Jinnia. And Meg, the priestess,
faced dual judgment as she brought Daiv,
her man, back to the tribe.

In these latter years of the 35th Century, Mount

Rushmore, with its colossal statues of
Washington, Jefferson, Lincoln and Theodore
Roosevelt was known to the superstitious
tribeswomen as "the Place of the Gods." From a
pilgrimage to this holy spot (where she had met
Daiv and learned the great secret that the
Ancient Ones were Men) Meg was returning.
("The Priestess Who Rebelled," Amazing Stories,
October, 1939.)—Ed.

[Transcriber's Note: This etext was produced from

Fantastic Adventures April 1940.
Extensive research did not uncover any evidence that
the U.S. copyright on this publication was renewed.]
Behind them the sun sank lazily, a huge, lop-sided ball of fire too-
bright-to-be-looked-at, and the gathering fingers of dusk stole softly
through the trees, casting wan shadows where they touched. The air
was thin and cold with the breath of approaching night; imps of
darkness lurked in the tree-roots and hollows.
But Meg felt no fear. She was alive with warmth, light, happiness.
The hills about her were soft-bosomed and gay with autumn's
garments; no longer was the landscape salt and drear as it had been
back there in the 'Kota territory, in the Place of the Gods. This was
her homeland, her native Jinnia. Beyond that turn was a rill, a half
day's march beyond the rill lay the village of her own
tribeswomen....
From her seat astride Nessa, she looked down upon Daiv, her Man,
and her voice was vibrant with happiness.
"We are almost there, Daiv! Soon you will meet my people, talk with
them, learn to love them as I do."
Daiv smiled at her dubiously.
"That I am prepared to do, Golden One. I can't help wondering,
though, how they will accept me. After what you've told me about
the Men of your tribe—" He shook his head.
Meg made a happy-sound deep in her throat; slipped from Nessa's
back and covered Daiv's lips with her hands.
"There is nothing to fear, Daiv. The Men of my Clan—pah!—they are
not like you. They are weak things, meant only for breeding. The
Mother will know, the moment she sees you, that you are one like
the Gods. She too made the pilgrimage. She will rejoice and be glad.
And—" Meg moved closer into the circle of his arms. "She will be
gladder still when she sees our happiness."
"I hope so," said Daiv soberly. His lips met hers in the touching-of-
mouths that he had taught her. Then, "But it is too late for us to
finish our journey today," he said. "We must find a place to make
camp."
"Just beyond the turn," Meg told him excitedly. "I know every inch of
this country, Daiv. When I was young, and studying under the
Mother to become a Priestess of the Clan, I used to ride out here to
seek solitude and the guidance of the Gods. Just beyond the turn
there is forest and a small stream. This is an ancient land, my lover;
the trees are mighty and strong. But—" Her eyes found his adoringly.
"But not so mighty and strong as you."
Daiv said, "Now, Golden One!" reprovingly, but Meg felt happy-
crinkles in her heart as she saw the way he drew back his shoulders
to lead the way into the forest. She did not mount Nessa again, but
walked behind Daiv, leading the tamed and captive doe by its bridle.

Nessa had been Daiv's gift to her, a mount to take the place of the
horse Meg had lost on her way to the Place of the Gods when the
Wild One had attacked her. "A wedding gift," Daiv had called it—
which did not make sense. But, then, Daiv was always saying
strange things. That was, Meg thought, a trifle awefully, because he
came from Kirki, which was a holy place far to the southland, near
the Land of the Escape. His tribespeople were direct descendants of
those Ancient Ones who, long ages ago, had fled from Earth to the
evening star in the bowels of a metal bird that spat fire.[1]
Daiv's skill had captured Nessa, wild woodland doe, but Meg's
gentleness had tamed her. She had borne Meg across all these
foreign territories; through Braska and Zurrie, to the blue-shining
fields of Tucky; now, at last, back to Meg's beloved Jinnia.
It had been a long journey and a strange one. Many things had Meg
seen; curiosities that would thrill the Women of her Clan to startled,
"Aiees!" of astonishment when she told of them. She had seen the
long, straight roads built by the Ancient Ones; their smooth creet
now cracked and worn but still easier to travel than the tortuous
woodland paths. She had seen the remnants of a gigantic hoam
called Sinnaty, where once had lived a great folk called "the Reds."
She and Daiv had lingered for three days with the tribeswomen of
Loovil (whose Mother knew the Mother of Meg's Clan, and had sent
greetings and gifts of ripe, fragrant bacca); there had they seen an
antique statue of a horse-god named Manowah.
They had even—Meg thrilled at the recollection—slept in one of the
traditional oaken hoams peculiar to the Tucky folk; dwellings
immortalized in the Tucky tribal song, "—sing one song of my oaken
Tucky hoam—"
And, "It is such a great land, Daiv!" Meg had said one night, lying
sleepless and excited over the wonders she had seen that day. "Such
a great land, this Tizathy! How I wish I had lived in it when the
Ancient Ones had welded it all together."
"A great land, indeed," was Daiv's answer. "But what did you call it?
Tizathy?"
"Yes," explained Meg. "That was its name. One of the ancient songs
tells of it. 'My country, Tizathy; sweet land of liberty—'"
Daiv looked at her with vast respect. "Someday we must visit my
people, Golden One. The elders of my tribe will want to talk to you.
You have such great learning...."
But now, at last, their journey was over. Jinnia's sweet green hills
cradled them; tomorrow they would join and touch hands with Meg's
sisters. Tonight they would sleep in the little forest Meg knew so
well....
Daiv turned, an unfamiliar wrinkle-look on his forehead. "Golden
One, didn't you say we were coming to forest land?"
"Yes, Daiv. Right before you. There—"
Then Meg was at his side, and her eyes were round with
wonderment. "But—but this is not right!" she cried.
It was no forest that confronted them. Where Meg's memory had
told her would be a riotous jungle of intertwining green, great trees
that brushed the heavens, high archways of leaves and thick-boled
woodland monarchs, there was nothing but a vast and desolate
plain, strewn with bristling twigs!
Stark and severe was that glade, swept bare of all vegetation save
these thousands upon countless thousands of twigs. No grass, no
shrubs, no flowers. Rough, bare hillside, and ankle-deep—the host
of stunted branches.

"I—I don't understand!" said Meg bewilderedly. She looked at Daiv,

fear suddenly cold in her heart, and she made a magic sign over her
breast to ward away the evil wood ghosts. "This is not as it should
be, Daiv! Something—"
The plain echoed Daiv's boisterous happy-sound.
"Something tells me," he chuckled, "you've made a mistake, Golden
One. So you know every inch of this country, eh? Well—" He
shrugged. "It is a cloudless night. And plain or forest, this is as good
a place as any to make camp. Get water, Meg, for the cawfee, and I
will build a fire."
Silently, with a sun-heat burning her cheeks, Meg moved to the rill
and got water. Then, as silently again, she returned to the spot Daiv
had designated. By this time she expected he would have made fire-
sparks with a rock and the bit of flame-metal he carried in his pouch
—but to her great surprise, no crackling blaze awaited her.
Instead, Daiv was standing upright beside one of the branched twigs
that festooned the plain. There was an angry-look on his face;
perspiration dripped from his forehead and his throat. The look he
turned to her was red with shame.
"Meg," he began, "Meg—a magic is upon me. I am weak. I have no
strength!"
"Strength, Daiv?"
"Yes. Look!" He bent to the broken branch before him. His strong
thighs tensed; the muscles of his back and shoulders corded with
strain. Fresh perspiration broke from his pores as he strained to lift
the tiny twig. Then his hands, white-knuckled and trembling, lifted
clear; he looked at her again.
"It is so small," he said in a faint, hurt voice. "Yet I cannot lift it!"
Meg sprang to his side; bent to the twig. She was slim, a pale,
golden shadow beside Daiv, but she was strong. Her hands grasped
the rough bark; she lifted—
And fell forward, thrown completely off balance by that weight
imponderable. Coarse soil rasped her knees, but she did not feel the
pain. All lesser emotions were lost in the swift, superstitious fear that
engulfed her.
"The forest is accursed, Daiv! We must flee!"
Hand in hand they raced wildly across the plain to the shelter of the
woodland at its farther rim. The rays of the dying sun cast their
shadows long before them, and a dry rattle of mirth seemed to rise
from the tangle of unyielding twigs that bruised their feet....

Meg dreamed fear-dreams that night. She was lost in a jungle of

trees hard as bone; as she fled beneath them, these trees groaned
and toppled toward her, their motionless branches clutching like
skeletal fingers. She moaned, cried Daiv's name—then wakened to
find him leaning over her anxiously.
"Meg! Listen!"
In the cool morning the sound carried clearly to Meg's ears. A
human voice, high-pitched in hideous screaming. A hoarse, grating
voice. Meg shuddered.
"A Wild One, Daiv! He has fallen into one of our traps."
"Wild One? Trap?"[2]
"I'll show you." Meg rose swiftly, instantly awake as any forest
animal. Bright morning sun cascaded down upon her, wakening a
sister gleam of gold on her arms and thighs, touching to warm life
the tawny down between her high, proud breasts. Save for the white
girdle of flesh beneath her fur breech-clout, Meg was all gold; her
hair, piled in a loose knot upon her head, was like a shining crown. It
was not all Women, Daiv thought briefly, whose charm withstood the
early morning sun. He was lucky to have found as a mate this slim
lance of loveliness.
He hungered for her lips. But he was Daiv—"He-who-would-learn"—
and here was a new mystery. He followed Meg. Meg followed the
plaintive cries.
They stopped, at length, at the lip of a cleft in the earth. It had
formerly been covered over with a webbing of boughs and ferns, but
now that cover was broken, and from the bottom of the pit came the
howls of pain that had drawn them.
Meg's lips were grim, white lines.
"He is in there," she said—and as she spoke she unslung her hunting
bow, slipped a bone-tipped arrow from her quiver. She stepped to
the mouth of the crevice, drew aim. Then—
"Wait, Golden One!"
Daiv swept the weapon from her hands. He looked down into the
pit, cried out sharply, then, ignoring Meg's warning, lowered himself
into it. A moment later he was back again, slipping his burden from
his shoulder. His burden was, as Meg had guessed, a brutish, hairy
Man-thing; foul with the stench of unwashed sweat and grease,
grimy with blood and dirt.
"You were going to kill him!" Daiv accused sternly. "He is a Man,
wounded, and you were going to kill him!"
The priestess said haughtily, "He is not a Man, he is one of the Wild
Ones. Of course I was going to kill him. That is the Law."
"It is a poor Law," grunted Daiv. He was bending over the Wild One
now, cleaning his wounds with handfuls of clean, dried grass. "If the
Women of your tribe build traps like these for Men, I'm not sure I
want to meet them. Aagh! False tops, and sharpened sticks
beneath!"
Meg the Priestess disappeared, and in her place stood Meg the wife,
a look of bafflement in her eyes.
"But, Daiv—" Faintly. "You killed one of the Wild Ones yourself. The
first time we met."
"That," said Daiv curtly, "was because he tried to linber you. I
wanted you for myself. There—he's coming around now. How do you
feel, Man? Are you all right?"
The Wild One's eyes were uncomprehending as they saw the
golden-limbed priestess and this strange, hairless Man before him.
His bearded lips parted in a strangled fragment of speech.
"I am ... all right." Then, to Daiv alone, "You ... saved my life!"
Daiv nodded. Thoughts crawled slowly behind the Wild One's eyes;
he reached a decision. From his filthy loin cloth he drew a chipped
and rusted blade; this he offered to Daiv. With the other hand he
smoothed flat the verminous tangle of hair above his heart.

"My life is yours, stranger," he said humbly.

Meg's eyes were wide with astonishment. From infancy she had
been taught that the Wild Ones were mad creatures without mercy,
without human sentiments; beasts that prowled the forests with but
two thoughts in their minds: to satisfy the hungers of their bellies
and to kill. Yet here was a Wild One displaying the civilized emotion
of gratitude. To Daiv she said querulously,
"He must be crazed, Daiv! Let us take him with us to the village. The
tribe Mother will want to see this marvel—a Wild One with a
Woman's instincts."
"He goes free!" said Daiv. He was still curt. He lifted the Wild One to
his feet. "Can you find your fellows, Wild One?" he asked.
The Wild One nodded mutely.
"Then, go!" ordered Daiv. "And be more careful of traps in the
future. Begone!"
But the Wild One hesitated an instant longer. The words came
haltingly from his lips—but they sprang from his heart. "My life is still
yours, Hairless One. Should ever you decide to claim it, you will find
me north of this spot. In a hillside cave by the waterfall...."
Then he was gone; a brutish, gnarled, hairy shadow sliding through
the matted jungle. And to Meg, "Come. Let us find your people,"
said Daiv. "I am minded to see what folk would harm poor brutes
like that one."
Meg pondered for hours, as they marched those last few miles to
her native village, but she could not quite discover why it was that
her cheeks and throat felt so hot. It was as though the fever-god
was within her, but she knew she was not ill....
And so, finally, she riding upon Nessa's back, Daiv striding before her
on swift, sure feet, they came to Meg's home. To the village of the
Jinnia Clan that was her own.
And again the remembrance of the massive twig and the strange
forest returned, bringing with it a half-fear. It lay uneasily in their
minds like the brooding residue of a dream....
 CHAPTER II
The Invaders

The glad word sprang first from the lips of the Warriors who guarded
the gates of the village. "It is Meg! Meg has returned from her
pilgrimage. Tell all!"
It spread to the Workers in the fields; they rose from their labors,
wiping grit-coarsened hands upon their thighs; their eyes
brightened. "Meg has returned!" And the breeding-mothers heard,
they lolling their plump, lush bodies in the sunlit doorways. They
heard, and their soft eyes filled with ready tears; they waddled
forward, their bulging hips swaying like ripened corn. "It is Meg,
come back from the Place of the Gods!"[3]
The Men, too, heard. They simpered foolishly and rolled their great,
soft eyes and primped their oiled hair. And word reached the tribal
Mother who came from her hoam to meet the Priestess. They met
within the confines of Meg's village; the aged Mother moved to greet
Meg with arms outstretched.
"You are returned safely, my child. May the Gods live forever!"
Meg could not speak for the little happy-imp who clutched her throat
and made water in her eyes. Her hands, tight and hot on the
withered hands of the old Mother, were speech enough. The Mother
bent forward and made a sign upon Meg's forehead.
"You have learned the Great Secret, my daughter; I can tell that by
the look in your eyes. Now you have passed the last barrier between
you and the Motherhood of our Clan. Tonight we will have a great
feast; at its conclusion I will invest you in the ultimate mysteries of
your task—"
There was a stir in the crowd surrounding Meg and the Mother; Daiv,
chafing with impatience at being thus ignored, had thrust himself
forward to his wife's side.
"What is this, Meg!" cried the Mother. "Have you turned Warrior as
well as Priestess? Where did you capture this hairless Man-thing?"
Now was the moment Meg had been at once dreading and looking
forward to. She placed her hand proudly within Daiv's, and her voice
was the clarion call of trumpets.
"It is no Man-thing, Mother. It is a Man; a real Man such as were the
Gods! Not a scrimping parody like our breeders, nor a foul brute like
the Wild Ones—but a Man. He is Daiv, my mate!"
"Mate!" The word leaped not only from the mouth of the Mother; it
was rasped by the Workers and the Warriors, it was piped in the
shrill, frightened quaver of the breeding-mothers. The Mother's eyes
clouded.
"Mate, Meg? What madness is this? Surely you know a priestess who
would be a Mother may not mate with a Man!"
Daiv said humbly, yet at the same time pridefully, "So Meg believed,
O Mother, until I taught her differently—and until she learned the
Great Secret at the feet of the Gods. I am Daiv, known as 'He-who-
would-learn'; I come from the place of the Escape. My people live by
the Laws of the Ancient Ones. In our land Man and Woman are
equal; we give and take love in the sacred customs of marriage."
Meg wished desperately that Daiv had said nothing. Given time to
lead up to these revelations, she felt she might have presented the
argument more convincingly. But in Daiv's deep voice these truths—
which she now recognized to be truths—sounded like rankest
heresies.
And they drew from the assembled listeners the response Meg had
feared. There was the snicking of metal upon metal as the Warriors
half-drew their swords from their scabbards; a low rumble of dissent
growled from the throats of the Workers. The breeding-mothers
squealed like stricken animals, fled with hands covering their ears
lest the Gods destroy them for having heard this impious outburst.
Lora, Chieftain of the Warriors, stepped forward, her lean jaw grim.
"Blasphemy, O Mother! By Tedhi, She-who-laughs, this Man-thing
befouls us with his lies. Shall I strike him down?"
She took a step toward Daiv. Meg cried out, moved between them,
turned beseeching eyes to the Mother. "No! I beg of you, O Mother,
no! Look upon Daiv! Look—and remember that which you saw, many
winters ago, in the Place of the Gods! You know I speak the truth,
Mother, and that Daiv, too, tells that which is so.
"Tell my sisters that this is well; that this is as it should be. You know
—"
Jain, Captain of the Workers, shook her head sorrowfully. In a gentle
voice she said, "Our priestess has gone mad, O Mother. The rigors of
the pilgrimage have been too much for her. What is the law? Death
for her, as well as for this hairless Man-thing? Or, having taken
herself a Man, must she become a breeding-mother?"
But the Mother stayed her. There was a faraway look in her eyes;
Meg knew that the aged leader of the Clan was remembering a
pilgrimage made many, many years ago to the Place of the Gods.
The Mother, Meg knew, had once looked upon the majestic figures of
Jarg, Ibrim, Taamuz and Tedhi on their great, rocky promontory at
Mount Rushmore and had seen, as Meg had seen, that the Gods
were, in truth, Men like Daiv. A word from the Mother now....
The Mother spoke. There was infinite sadness in her voice.
"It is the Law," she said, "that none shall strive to change the ways
of the Clan. You, Meg, have ignored the Law. You and your mate will
be given justice."
And she turned away.
A gasp spun Meg's gaze to Daiv. His face was crimson with an
anger-look; great veins throbbed in his forehead. He roared, "Here,
then, is the joyous welcome your Clan offers us, Golden One!
Justice? What kind of justice may we expect from a doddering,
thwarted old harridan—"
"Daiv!" screamed Meg.
But her cry broke too late. With one swift gesture Daiv had
wrenched the sword from the hands of Lora. Now he tested its
blade, swept Meg into the circle of his arms, and laughed at the
startled clanswomen defiantly.
"So you would judge a Man?" he taunted. "A Man of the Kirki tribe?
Come, then, you filthy diggers of dirt and loveless scarecrows. Let
your judgment be the matching of my steel against yours!"
There was a tense moment of silence. Then anger, bitter as the fruit
of the simmon tree, flamed in the voices of Meg's sisters. A score of
Warriors sprang forward, swords drawn. At their flank advanced the
Workers, hoes and adzes uplifted. Meg smiled piteously at Daiv and
murmured a swift prayer to the Gods. It was grievous to die thus,
before the blades of loved ones....

And a faint, thin cry stayed them all! They turned to see, at the
deserted southward gate, the torn and bleeding figure of a Warrior
who, hair disheveled, face scarred and raw, hands and arms deep-
scored with gory cicatrices, pulled herself within the Jinnia camp
dragging behind her one sturdy leg and one blackened, withered
stump.
In that moment of dread wonder it did not even seem strange to the
clanswomen that the first to reach the wounded Warrior's side
should be the stranger, Daiv. But Daiv it was who raised her in his
arms.
The visitor's eyes were filmed with pain, horror, fatigue. They
unveiled now, and an indomitable purpose shone through. In a
husky voice she faltered, "It is too late ... to save me. Soon I will
join ... my Clanswomen ... and the Gods. Save ... yourselves!"
There was unbelievable gentleness in Daiv's voice.
"What is it, Warrior?" he asked. "What enemy thus cruelly destroyed
you? Of what would you warn us?"
From some deep-hidden well the messenger drew new strength. Her
eyes blazed as she answered, "I bid you flee to the secret spots of
the mountains. An evil foe even now marches upon your camp.
Stunted and vicious little yellow-skinned Men-things who linber[4]
our Clans, destroy our fighters with tubes that maim and stun."
The aged Mother was beside her now.
"Who are you, daughter?" she begged. "Whence come you?"
"I am Vivyun," labored the refugee, "of the Durm Clan. Short days
ago came strange lightings in the heavens; mad thunders burst in
the forests about our village—"
Jain interrupted, startled, "Mother! The omens we heard night
before last in the forest to our west!" and Meg looked swiftly at Daiv.
She cried,
"The forest through which we fled, Daiv! The wood of heavy twigs!"
Daiv silenced her with a thoughtful nod. Vivyun's halting speech
continued.
"—then came the onslaught. Armored demons, the color of mustard
seed, burst upon us. Our Warriors went to meet them but the dwarfs
loosed lights from sticks and where the soldiers had stood, now
were but inch-long, stony parodies of Women. One of the lights
played for an instant upon my leg—"
Meg looked and shuddered. The dying Warrior's leg was firm and
round from hip to thigh; ten inches above the knee it ended abruptly
in a scoriated stump from which depended an ugly, wartlike
excrescence which—Meg saw with sickening horror—was the
perfectly formed simulacrum of a human limb.
Daiv was muttering savagely, "Speak on, Warrior!"
"They come," persisted Vivyun, "to capture Women. Like the Wild
Ones, they die out for lack of Mates. Out of the far southland they
come, from a land called Mayco. They bear other strange weapons.
A stick that shoots lights of insanity ... a wall they build of invisible
bricks...."
"More!" pleaded Daiv and the Mother in one breath as Vivyun
faltered. "More!"
But a strange, foolish look glazed the dying one's eyes. Her lips
moved whitely and her breath was a whisper. "You are a strange ...
creature," she said to Daiv. "Somehow you ... make it easy to die ...
Man-thing...."
Then she was still.

Lora, Chieftain of the Warriors, broke the spell that bound them all
with a thunderous cry.
"Invaders? No invaders can take the village of the Jinnia tribe! To
arms, Warriors! To your posts. Let these yellow dwarfs attack us,
and—" She laughed evilly.
Daiv sprang to his feet; his voice a peremptory challenge. "Hold,
Warrior! Did you not hear what Vivyun said? These invaders have
magic weapons; sticks that spit insanity and crumpling death. It is
best we should flee to the hills. Maybe there we can devise some
way—"
Meg's cheeks were hot with sorrow for Daiv as the Warrior Chieftain
scorned him with her eyes.
"It is a Man-thing after all!" she spat. "A hairless Wild One with the
cowardly instincts of all Men. Fool! Know you not the dying one
babbled foolishness in her delirium? Sticks that dwarf Warriors! Walls
without bricks!"
Daiv gritted, "I have no time for argument, Warrior." To the Mother
he cried beseechingly, "There is little love lost between me and thy
Clan, O Mother. But because you are Meg's sisters, I would see you
live. Believe me, there was truth in Vivyun's warning. I myself have
heard elders speak of a sunlit land called 'Mayco,' peopled by savage
demons—"
The Mother pressed her hands together in an agony of indecision. To
Meg, in her desperation, she turned, crying, "See now, O my
daughter, how heavy is the task of being a Mother?" And she
muttered, half to herself, "If this be true, then all are doomed unless
we flee. But if it be lies—"
Daiv, man of action, tired swiftly of this maundering. For the second
time that afternoon he reached for Meg's hand.
"Come, Golden One! Let these fools die; let them become stiffened
twigs of humans as the branches we saw in the forest were stiffened
and dwarfed trees! I will take you to safety—"
Meg took a step forward. And then—one of the Women laughed. A
sneering laugh. Meg's cheeks flamed, and her outstretched hand
dropped to her side. She shook her head.
"No, Daiv." Sadly. "I had not dreamed you were—"
"A coward?" Daiv supplied the word wrathfully. "I am a coward to
wisely flee from the magic of men who know the secrets of the
Ancient Ones? By the Gods, Golden One, it is you who have lost your
senses. If you will not come willingly, I'll save you in spite of
yourself. Come!"
And he sprang toward her. Meg stumbled backward, torn by a
thousand conflicting emotions. Then, of a sudden, came that which
coalesced all her emotions into one indistinguishable chaos. There
came a mighty roaring sough from the woodlands south of the
village; a portion of the walls caved inward with a mighty crash;
spent air howled like the breath of the flood-time gods, and—
In the opening, golden sunlight gleamed on glinting armor! A horde
of dwarfed and evil yellow men, shining sticks in their hands,
stormed in through the rent!

CHAPTER III
The Mate of Grensu

What happened then was never afterward to be clear in Meg's mind.

She realized that the air was alive with the cries of the attacking
dwarfs; that these cries found echoes in the shouts of her Warrior
sisters who sprang forward to meet them.
She was conscious that a Warrior at her side, with a half-uttered
choke of fear, had suddenly met breast-high a streaming light
expelled from one of the invaders' sticks; she heard the clatter of
metal upon rock as the Warrior's sword fell. She did not realize she
had stooped instinctively to retrieve the fallen weapon until she
found herself charging forward, cries ripping her throat, the sword
waving above her head.
There seemed to be two Megs; one who raced futilely, vaingloriously,
toward that crouching, smirking band of attackers; the other who
stood somewhere apart from the mad press, watching the battle
with impartial judgment. It was the first Meg who flashed down
upon a stunted yellow man unnoticed, swung her heavy sword in a
flailing motion that split his hauberk and sent his headless body
toppling to earth.
It was the second Meg who noticed, with incredibly cool appraisal,
that from the sticks of the invaders emerged two different types of
light. One, a pale, greenish light, caused those bathed in it to drop
their swords, cease their shouting, wander aimlessly off across the
blood-drenched field. The other, a cherry-flamed light, was the
horror of which the Durm Warrior, Vivyun, had warned.
The attackers seemed only to use it when dire necessity pressed. Its
results were ghastly. Meg's brain reeled before the shock of seeing a
Worker on her right run full-tilt into that cherry beam. One instant
the Worker was there; the next she was gone. A sharpened hoe lay
beside a blasted doll-like thing from which, momentarily, rose a
steamy mist and a nauseating stench.
Given weapons to match those of the yellow dwarfs the Jinnians
might have won through. Their numbers were as great as those of
the invaders; their spirit was that of Women fighting for their native
homeland. Gallantly they pressed onward, forward—and as gallantly
they died. Save for that greater portion of them who assumed the
"life-in-death" Meg had marked; the stupid insanity that sent them
staggering, weaponless, upon mindless errands.
For conquerors, the yellow men waged stupid war. They seemed
more intent on capturing prisoners than on destroying—or perhaps
they had not anticipated such a stubborn resistance. Howsoever that
may have been, time and again a member of the Jinnia Clan,
evading the sticks-that-flamed, would pierce the enemy lines. There,
ere the cherry light steamed her body into brittle stone, her sword
would draw the life-blood of a yellow invader.

Meg had learned much in her long pilgrimage to and from the Place
of the Gods. Daiv had taught her how to take advantage of all
natural protections when warring against a superior force. These
guerilla tactics served her well now. With the first conflict of forces
she had sprung to a place of concealment behind the ruptured wall;
from this vantage point she could see straggling invaders as they
entered the village; could not be seen by them until their eyes
widened at the sight of a dripping sword thirsting for their throats.
Four died thus beneath her blade. Cautiously, now, she ventured a
glance into the yellow men's defense line.
There she saw what her quick intelligence told her must be the
object of her attack. Outside the village stood a tiny knot of dwarfs
garbed in armor more glittering, more ornate, than that of those
who made the attack.
These, Meg recognized, were the leaders. The commander-in-chief
must be that overripe, ochre plum in golden greaves and casque; he
who stood impatiently fingering the handle of his light-stick as he
watched his warriors' progress.
To think was to act. It never occurred to Meg that her solo foray was
suicidal. Hurdling the bodies of those before her, she leaped through
the broken wall; raced, bobbing and weaving, shifting her course to
make herself an impossible target, down upon the commander's
party.
As she ran, her hair broke loose from its handknit snood; lithe
muscles snapped the sinews that held her cloak. She was like some
magnificent golden panther as, hair flowing behind her in a liquid
honey stream, high, firm breasts rising with the quickening of her
breath, she charged down on her tribe's enemies.
Thoughts flashed dizzily through her mind. A great burst of
exultation; she was too near, now, for them to stop her! Then a
soul-shaking disappointment. She had been seen! One of the
officers' eyes bugged; he raised a light-stick—
Then most incredible thing of all—the commander-in-chief had seen
her, and his porcine eyes, slanted and deep-sunk in rolls of saffron
flesh, were glittering with delight. His left hand was beating down
the cherry-flame of the lieutenant as his right was pointing at her
breast his own stick. Light flashed—pale green. Something within
Meg seemed to snap; suddenly she was suffused with a sense of
coolness, a bewildering drainage of the fever that had coursed
through her veins.
So funny. So funny to have thought this battle important. It wasn't,
really. It was all a mistake. And the sword in her hand? Meg glanced
at it idly, her charge slowing to a walk. She cast the sword away.
The din of conflict was a thin and distant sound. The world about
her was sweet and green ... the clouds billowed on an endless blue
like boat-sails scudding before the wind. There was something she
should remember. There was a dancing haze before her eyes ...
flowers about her feet. Were she to wander gently, now, to that
farthest field—what was it she could not remember?—there would
be golden buttercups and the prim, starch cornflower ripe for
plucking.
Her body was numbed and drowsy with a sense of comfort. Only—
there was a Man; a Man named Daiv—only she could not be happy
here. Not unless she forgot her troubles, forgot the man named
Daiv, forgot the world was spinning and reeling and swirling before
her eyes like a gigantic wheel going faster and faster and faster....
Then there was blackness.

Her first thought was that she had fallen, momentarily stunned, on
the field of battle. She woke with a start, groping for the sword that
should be by her side.
It wasn't there. She touched the flabby flesh of a breeding-mother
who, flaccid-breasted and aquiver, shook beside her in an ecstasy of
fear. Meg gagged as she stumbled to her feet. Her limbs were still
weak beneath her, as if the veins that fed them had been fouled; her
head was filled with tiny imps who danced and shrieked unmercifully.
But—she was alive! And the mists were clearing from her brain.
Now she knew there was sobbing beside her. Strange sobbing. Not
the soft, easy gulping of a breeding-mother; a harsh sound like the
rasp of an adze on creet. It was Lora, Chieftain of the Warriors. Her
armor crusted with blood, her great hands twisting with grief, she
was rocking backward and forward, alternately weeping and cursing
the Gods.
"Now accursed be the breeding-mother that gave me birth!" was her
plaint. "This night shall my stars burn as cinders—"
Meg shook her shoulder roughly.
"Lora!"
The Warrior Chieftain's eyes recognized her. Lora cried prayerfully,
"Search well your girdle, Meg! Have you a dagger upon you?"
"No. But, why—?"
Lora beat her tiny, thwarted breasts with clenched fists. "I live!" she
choked. "I, their leader, continue to live, while they lie there, in
peace and glory—"
Meg saw, then, that she was part of a group huddled in the center
of that which had been the fortress of their Clan. They numbered
more than three score; a mixed group of battle-grimed Warriors,
Workers, breeding-mothers, even one or two pale-faced, weeping
Men. The studs of the Jinnia Clan.
But there was another group at the far end of the court. These
would never again either laugh or mourn. They were the dead.
Workers and Warriors for the most part, although a few plump,
bulbous bodies fed the mound. In still another place lay the bodies
of the slain invaders; these had been accorded more dignity. All
about the arena lay curiously shaped pebbles which—Meg knew,
shivering clammily—were not pebbles. Two stunted yellow men,
grinning callously, now busied themselves raking up these grisly
objects.
Meg said, "The—the Mother?" and as if in answer to her thought, a
gentle voice reached her ears.
"I am here, Meg, my daughter."
Meg turned swiftly. The Mother of the Clan lay behind her,
motionless, head lifted upon a bolt of cloth someone had provided.
There was an image of dreadful pain in the Mother's eyes. Meg
sprang to her side, heart bursting with sorrow.
"Mother—you are hurt!"
"Nay, daughter, I am slain." The Mother sighed; a wan breath of
regret. "They had no intent to kill me. But the rays were too potent
for my aged body. I will linger yet a little while, then I must go. It is
sad that I must leave my Clan captive to a race of beasts like these."
Meg said, "Rays, Mother?"
"Yes, my child. Those weapons which our Warriors could not
comprehend are similar to those which, in the old legends, it is told
the Ancient Ones used to destroy each other. Vibrations that cause,
in one case horrible death; in the other case, stupefaction."
"But—but how?"
"I am not sure. But I think the cherry-light has the power of
absorbing all water from the human body, thus dwindling it instantly
to a husk. The green ray interrupts the nerve-centers, breaking the
brain's contacts with other organs." The Mother's face was resigned.
"All these things and many others you would have learned, Meg, had
not this catastrophe come upon us. And had not you returned from
the Place of the Gods with—a mate."
Her last word reminded Meg of that latent question which the green
ray had driven from her mind. Instantly her heart was cold with
dread.
"Daiv! Oh, Mother—Daiv! I do not see him. Is he one of—of those?"
Her eyes stared with horrid fascination at the tiny pile of simulacra
the invading warriors were raking together. But it was not the
Mother who answered. It was Lora, now come to their side. The
Warrior Chieftain's lean face was etched with scorn.
"No, Meg, your Man-thing is not there! That would be an honorable
death for him."
Meg faltered, "Then—then where—?"
The Mother's eyes, in pity, would not meet Meg's. "He fled, Meg."
"Fled! Daiv fled!" Meg stared, every fibre of her body taut as the gut
of a bow. "I—I do not believe it!"
Even Lora's harsh voice was more gentle as she said, "It is true. He
was but a Man, Meg; a Man, weak and cowardly. With the first
breath of fighting he turned and fled the camp. Into the hills
beyond."
The Mother intervened, "Perhaps it is better so, my daughter.
Perhaps your own madness escaped on his limbs? If ever we win
free of this host—"
Meg's new-found world of love and happiness crumpled into shards
about her feet. There was a redness on her cheeks, for Daiv and for
herself who had let his mouth touch hers. The rains of weakness
filled her eyes, and she said, "So be it, Mother—"

It was at that moment a coarse, guttural voice interrupted her.

"Ah—there she is! That's the one, Leekno. Step forth, ivory one!"
Meg turned. Staring at her, his lips red and moist as the pepper-fruit,
was the golden-greaved and helmeted commander whose actions
had at once saved her life and broken her will for battle. A fungoid
puff-ball of a man with twisting mouth, his eyes upon her made Meg
feel suddenly naked and unclean. She shuddered.
"Step forth, I say!" repeated the leader. A yellow soldier moved to
enforce his command. Meg shook free of the underling's touch,
moved a step forward with a proud freedom of motion that wakened
a dancing light in her accoster's eyes.
His voice was a purr of satisfaction.
"I was right. Very well, Leekno, you may distribute the other captives
to our men by lot. And mind there be no wrangling amongst them.
This barbarian I will take to my tent."
Meg demanded, "What is the meaning of this, little monster? Who
are you to thus address a Priestess of the Jinnia Clan, and what
would you of me?"
The commander's slant eyes blinked in appreciative delight. "A
spirited filly, this!" he murmured. "Know, then, Woman, that you are
greatly honored. I am Grensu, captain of this band. We are a legion
of the mighty race of Japcans who rule in the sunny land of Mayco,
many days' march to the south of here."
From behind Meg the aged Mother's voice stirred in sighing wonder.
"Mayco? In the books of the Ancient Ones it is written of a land by
that name. But its rulers were white men—"
Grensu made mirthless happy-sounds in his throat.
"The old one surprises me. Aye, withered crone, in the early days it
was as you say. But that was before the great wars, and before the
rebellion of the Women. Even then there were in Mayco many of our
race, children of the Sun.
"But when mankind destroyed itself and the Ancient Ones died out,
battling first each other, then between their sexes, we stayed aloof;
we waited and planned and bred. Our Japni blood mixed well with
that of the Maycans, giving rise to the mighty stock we now
represent—the Japcans." Grensu glanced down his own obese frame
pridefully. "Now, not only are we the possessors of the secrets of the
Ancient Ones' war-weapons, but we are become perfect in body and
brain."
Meg laughed scornfully, "Little fat-bellied lemon, I could squeeze
between my fingers—so!" And she took a step toward Grensu. For a
second he looked startled; then an expression of mingled pique and
admiration mottled his pudgy features.
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