Download Full Version ebook - Visit ebooknice.

com

(Ebook) Modern Algorithms for Image Processing:

Computer Imagery by Example Using C# by Vladimir
Kovalevsky ISBN 9781484242360, 148424236X

https://ebooknice.com/product/modern-algorithms-for-image-
processing-computer-imagery-by-example-using-c-7282956

Click the button below to download

DOWLOAD EBOOK

Discover More Ebook - Explore Now at ebooknice.com

 Instant digital products (PDF, ePub, MOBI) ready for you
Download now and discover formats that fit your needs...

Start reading on any device today!

(Ebook) Algorithms for image processing and computer

vision by J. R. Parker ISBN 9780470643853, 9781118019627,
9781118021880, 9781118021897, 0470643854, 1118019628,
1118021886, 1118021894
https://ebooknice.com/product/algorithms-for-image-processing-and-
computer-vision-2623344

ebooknice.com

(Ebook) Digital Image Processing Using Matlab by Rafael C.

Gonzalez ISBN 9780070702622, 0070702624

https://ebooknice.com/product/digital-image-processing-using-
matlab-10945078

ebooknice.com

(Ebook) Digital Media Processing: DSP Algorithms Using C

by Hazarathaiah Malepati ISBN 9781856176781, 1856176789

https://ebooknice.com/product/digital-media-processing-dsp-algorithms-
using-c-1537702

ebooknice.com

(Ebook) Digital Image Processing Using MATLAB (R) by

Rafael C. Gonzalez ISBN 9780130385604, 0130385603

https://ebooknice.com/product/digital-image-processing-using-
matlab-r-1273030

ebooknice.com
(Ebook) Essential algorithms : a practical approach to
computer algorithms using Python and C# by Rod Stephens
ISBN 9781119575993, 1119575990
https://ebooknice.com/product/essential-algorithms-a-practical-
approach-to-computer-algorithms-using-python-and-c-12274474

ebooknice.com

(Ebook) Essential Algorithms: A Practical Approach to

Computer Algorithms Using Python and C#- Second Edition by
Rod Stephens ISBN 9781119575993, 1119575990
https://ebooknice.com/product/essential-algorithms-a-practical-
approach-to-computer-algorithms-using-python-and-c-second-
edition-57280052
ebooknice.com

(Ebook) Fuzzy Logic for Image Processing: A Gentle

Introduction Using Java (SpringerBriefs in Electrical and
Computer Engineering) by Caponetti, Laura, Castellano,
Giovanna ISBN 9783319441283, 3319441280
https://ebooknice.com/product/fuzzy-logic-for-image-processing-a-
gentle-introduction-using-java-springerbriefs-in-electrical-and-
computer-engineering-55589264
ebooknice.com

(Ebook) Visual Perception Through Video Imagery (Digital

Signal and Image Processing) by Michel Dhome ISBN
9781848210165, 1848210167
https://ebooknice.com/product/visual-perception-through-video-imagery-
digital-signal-and-image-processing-1708444

ebooknice.com

(Ebook) Computer Vision and Image Analysis; Digital Image

Processing and Analysis by Scott E Umbaugh

https://ebooknice.com/product/computer-vision-and-image-analysis-
digital-image-processing-and-analysis-55456502

ebooknice.com
Modern
Algorithms for
Image Processing
Computer Imagery by Example Using C#
—
Vladimir Kovalevsky
 Modern Algorithms for
Image Processing
Computer Imagery by Example
Using C#

Vladimir Kovalevsky
Modern Algorithms for Image Processing: Computer Imagery by Example Using C#
Vladimir Kovalevsky
Berlin, Germany

ISBN-13 (pbk): 978-1-4842-4236-0 ISBN-13 (electronic): 978-1-4842-4237-7

https://doi.org/10.1007/978-1-4842-4237-7
Library of Congress Control Number: 2018965475

Copyright © 2019 by Vladimir Kovalevsky

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.
Trademarked names, logos, and images may appear in this book. Rather than use a trademark symbol with
every occurrence of a trademarked name, logo, or image we use the names, logos, and images only in an
editorial fashion and to the benefit of the trademark owner, with no intention of infringement of the
trademark.
The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not
identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to
proprietary rights.
While the advice and information in this book are believed to be true and accurate at the date of publication,
neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or
omissions that may be made. The publisher makes no warranty, express or implied, with respect to the
material contained herein.
Managing Director, Apress Media LLC: Welmoed Spahr
Acquisitions Editor: Joan Murray
Development Editor: Laura Berendson
Coordinating Editor: Jill Balzano
Cover image designed by Freepik (www.freepik.com)
Distributed to the book trade worldwide by Springer Science+Business Media New York, 233 Spring Street,
6th Floor, New York, NY 10013. Phone 1-800-SPRINGER, fax (201) 348-4505, e-mail orders-ny@springer-
sbm.com, or visit www.springeronline.com. Apress Media, LLC is a California LLC and the sole member
(owner) is Springer Science + Business Media Finance Inc (SSBM Finance Inc). SSBM Finance Inc is a
Delaware corporation.
For information on translations, please e-mail rights@apress.com, or visit http://www.apress.com/
rights-permissions.
Apress titles may be purchased in bulk for academic, corporate, or promotional use. eBook versions and
licenses are also available for most titles. For more information, reference our Print and eBook Bulk Sales
web page at http://www.apress.com/bulk-sales.
Any source code or other supplementary material referenced by the author in this book is available to
readers on GitHub via the book's product page, located at www.apress.com/9781484242360. For more
detailed information, please visit http://www.apress.com/source-code.
Printed on acid-free paper
Dedicated to my wife, Dr. Baerbel Kovalevsky
Table of Contents
About the Author����������������������������������������������������������������������������������������������������� ix

Acknowledgments��������������������������������������������������������������������������������������������������� xi
Introduction����������������������������������������������������������������������������������������������������������� xiii

Part I: Image Processing�������������������������������������������������������������������������������� 1

Chapter 1: Introduction�������������������������������������������������������������������������������������������� 3

Chapter 2: Noise Reduction�������������������������������������������������������������������������������������� 5

The Simplest Filter������������������������������������������������������������������������������������������������������������������������ 6
The Simplest Averaging Filter������������������������������������������������������������������������������������������������������� 6
The Fast Averaging Filter�������������������������������������������������������������������������������������������������������������� 8
The Fast Gaussian Filter������������������������������������������������������������������������������������������������������������� 14
The Median Filter������������������������������������������������������������������������������������������������������������������������ 17
Sigma Filter: The Most Efficient One������������������������������������������������������������������������������������������� 18
Suppression of Impulse Noise���������������������������������������������������������������������������������������������������� 23

Chapter 3: Contrast Enhancement�������������������������������������������������������������������������� 43

Automatic Linear Contrast Enhancement����������������������������������������������������������������������������������� 43
Histogram Equalization��������������������������������������������������������������������������������������������������������������� 45
Measuring the Lightness of Color Images����������������������������������������������������������������������������������� 49
Contrast of Color Images������������������������������������������������������������������������������������������������������������� 52
Manually Controlled Contrast Enhancement������������������������������������������������������������������������������� 53

Chapter 4: Shading Correction with Thresholding�������������������������������������������������� 65

Thresholding the Images������������������������������������������������������������������������������������������������������������ 75

Chapter 5: Project WFshadBinImpulse������������������������������������������������������������������� 81

v
Table of Contents

Part II: Image Analysis��������������������������������������������������������������������������������� 85

Chapter 6: Edge Detection�������������������������������������������������������������������������������������� 87
Laplacian Operator���������������������������������������������������������������������������������������������������������������������� 87
The Method of Zero Crossing������������������������������������������������������������������������������������������������������ 89
Are Zero Crossings of Laplacian Closed Curves?������������������������������������������������������������������������ 89
How to Eliminate Irrelevant Crossings���������������������������������������������������������������������������������������� 91
Noise Reduction Before Using the Laplacian������������������������������������������������������������������������������ 92
Blur During the Digitization and Extreme Value Filter����������������������������������������������������������������� 93
Fundamental Errors of the Method of Zero Crossing in the Laplacian���������������������������������������� 98

Chapter 7: A New Method of Edge Detection�������������������������������������������������������� 101

Means for Encoding the Edges������������������������������������������������������������������������������������������������� 102
The Idea of an Abstract Cell Complex��������������������������������������������������������������������������������������� 103
A Simple Method of Encoding Edges���������������������������������������������������������������������������������������� 105
Improvements of the Method of Binarized Gradient����������������������������������������������������������������� 107
Further Improvements of the Method of Binarized Gradient����������������������������������������������������� 120
The Edge Detector of Canny������������������������������������������������������������������������������������������������������ 122
Edges in Color Images�������������������������������������������������������������������������������������������������������������� 123
Conclusions������������������������������������������������������������������������������������������������������������������������������� 125

Chapter 8: A New Method of Image Compression������������������������������������������������ 127

Using a Cell Complex for the Encoding of Boundaries�������������������������������������������������������������� 128
Description of the Project WFcompressPal������������������������������������������������������������������������������� 131
The Project WFrestoreLin���������������������������������������������������������������������������������������������������������� 150

Chapter 9: Image Segmentation and Connected Components������������������������������ 167

Segmentation by Quantizing the Colors������������������������������������������������������������������������������������ 168
Connected Components������������������������������������������������������������������������������������������������������������ 168
The Graph Traversal Algorithm and Its Code����������������������������������������������������������������������������� 171
The Pseudo-Code of the Breadth-First Algorithm���������������������������������������������������������������� 172

vi
 Table of Contents

The Approach of Equivalence Classes�������������������������������������������������������������������������������������� 173

The Pseudo-Code of the Root Algorithm����������������������������������������������������������������������������������� 177
The Project WFsegmentAndComp��������������������������������������������������������������������������������������������� 179
Conclusion�������������������������������������������������������������������������������������������������������������������������������� 186

Chapter 10: Straightening Photos of Paintings���������������������������������������������������� 187

The Principle of Straightening�������������������������������������������������������������������������������������������������� 189
Codes of Most Important Methods�������������������������������������������������������������������������������������������� 196
Conclusion�������������������������������������������������������������������������������������������������������������������������������� 203

Chapter 11: Polygonal Approximation of Region Boundaries and Edges������������� 205

The Problem of Polygonal Approximation��������������������������������������������������������������������������������� 205
Schlesinger’s Measure of Similarity of Curves������������������������������������������������������������������������� 206
Statement of the Approximation Problem��������������������������������������������������������������������������� 207
Algorithms for Polygonal Approximation����������������������������������������������������������������������������������� 207
The Split-and-Merge Method���������������������������������������������������������������������������������������������� 208
The Sector Method�������������������������������������������������������������������������������������������������������������� 209
The Improvement of the Sector Method������������������������������������������������������������������������������ 210
Replacing Polygons by Sequences of Arcs and Straight Lines������������������������������������������������� 211
Definitions and the Problem Statement������������������������������������������������������������������������������� 211
The Approximate Solution��������������������������������������������������������������������������������������������������� 212
The Project WFpolyArc�������������������������������������������������������������������������������������������������������������� 217
Methods Used in the Project WFpolyArc������������������������������������������������������������������������������ 218
Precision of the Calculation of the Radii����������������������������������������������������������������������������������� 225
Conclusion�������������������������������������������������������������������������������������������������������������������������������� 226

Chapter 12: Recognition and Measurement of Circular Objects��������������������������� 227

Mathematical Foundation of the Method���������������������������������������������������������������������������������� 228
The Project WFcircleReco��������������������������������������������������������������������������������������������������������� 232
The Form of the Project WFcircleReco�������������������������������������������������������������������������������� 233

vii
Table of Contents

Chapter 13: Recognition of Bicycles in Traffic������������������������������������������������������ 243

Mathematical Foundation of Ellipse Recognition���������������������������������������������������������������������� 243
The Project WFellipseBike��������������������������������������������������������������������������������������������������������� 247
Another Method of Recognizing the Direction�������������������������������������������������������������������������� 258

Chapter 14: A Computer Model of Cell Differentiation������������������������������������������ 261

Conclusion�������������������������������������������������������������������������������������������������������������������������������� 266

References������������������������������������������������������������������������������������������������������������ 267

Index��������������������������������������������������������������������������������������������������������������������� 269

viii
About the Author
Vladimir Kovalevsky received his diploma in physics from the Kharkov University
(Ukraine), his first doctoral degree in technical sciences from the Central Institute of
Metrology (Leningrad), and his second doctoral degree in computer science from the
Institute of Cybernetics of the Academy of Sciences of the Ukraine (Kiev) where he
headed the Department of Pattern Recognition for more than a decade.
Vladimir has been living in Germany since 1983. He was a researcher at the Central
Institute of Cybernetics of the Academy of Sciences of the GDR, Berlin, a professor
of computer science at the University of Applied Sciences Berlin, and a scientific
collaborator at the University of Rostock.
He has been a visiting researcher at the University of Pennsylvania, a professor at
the Manukau Institute of Technology in New Zealand, and a professor at the Chonbuk
National University in South Korea. He has reviewed for the journals Applied General
Topology, Computer Vision and Image Understanding, IEEE Transactions on Pattern
Analysis and Machine Intelligence, and others.
Vladimir has been a plenary speaker at conferences in Europe, the United States, and
New Zealand. His research interests include digital geometry, digital topology, computer
vision, image processing, and pattern recognition. He has published four monographs
and more than 180 journal and conference papers on image analysis, digital geometry,
and digital topology.

ix
Acknowledgments
I wish to acknowledge valuable and fruitful discussions with Boris Flach, Reinhard
Klette, Ulrich Koethe, Alexander Kovalevsky, Volkmar Miszalok, and Peer Stelldinger.
These discussions have significantly contributed to this work.
I would like to express my special appreciation to Alexander V. Kovalevsky, who
helped significantly as an experienced programmer in the development of my projects.

xi
Visit https://ebooknice.com to
discover a wide range of
eBooks across various genres.
Enjoy exclusive deals and
discounts to enhance your
reading experience. Start your
digital reading journey today!
Introduction
This book presents a collection of algorithms and projects for processing two-­dimensional
images. I developed and investigated the algorithms. Special emphasis is placed on
computer solutions of problems related to the improvement of the quality of images,
with image analysis and recognition of some geometrically definable objects. New data
structures useful for image analysis are presented. The description of all algorithms
contains examples of source code in the C# programming language. Descriptions of
projects contain source code that can be used by readers.
With this book I intend to help you develop efficient software for processing
two-­dimensional images. There are a lot of books on image processing, but important
algorithms are missing from these books. I have developed many efficient algorithms as
a new and important contribution to this area.
I have paid great attention to solutions of problems in image analysis. On the other
hand, problems of improving the quality of images are important for the arts. My wife is
a recognized specialist in the history of the arts, and her publications often use copies
of famous pictures and drawings. The photographs of these artworks are often of low
quality. Often photographs of historical drawings illustrating the work of a painter are
of such low quality that it is almost impossible to clearly see the contents of the image.
Improving these images is therefore very important. In such cases, the programs I have
developed for improving the quality of pictures are very useful.
I have developed efficient algorithms for recognizing circles and ellipses in
noisy images. These algorithms can be used for recognizing objects with a shape
approximating a circle; for example, apples, mushrooms, and so on. They can also be
used for recognizing bicycles in images of traffic because the wheels of bicycles are ideal
circles, but if the bicycle is positioned in such a way that the plane of its frame is not
orthogonal to the viewing ray, then its wheels look like ellipses rather than circles. I was
therefore forced to develop efficient algorithms for recognizing ellipses in noisy images
as well. My efforts were successful and the book contains a chapter devoted to the
recognition of bicycles in noisy images.
The book contains descriptions of numerous algorithms for image analysis,
including these:

xiii
Introduction

• Manually controlled thresholding of shading corrected images.

• A fast algorithm for simultaneously labeling all connected

components in a segmented image.

• A new efficient method of edge detection.

• A fast algorithm for approximating digital curves by polygons and for

estimating the curvature of circular arcs approximating the curve.

• Algorithms for recognition and measurement of circular or elliptical

objects in color images.

Among the algorithms for image improvement, the most important are the following:

• The algorithm for rectifying photographs of paintings taken at an

oblique angle.

• An algorithm correcting images of nonuniformly illuminated scenes.

• The algorithm for improving the contrast of images of nonuniformly

illuminated scenes.

• The best algorithm for reducing Gaussian noise (the so-called

Sigma-­Filter).

• The algorithm for reducing impulse noise.

All descriptions are followed by a pseudo-code similar to the C# programming

language. Most of the descriptions contain source code that can be copied from the text
and used directly in a Windows Forms program written in the C# .NET language.
All source code and figures are included in a download file (which you can access via
the Download Source Code button located at www.apress.com/9781484242360) so you
can see the colors.

xiv
PART I

Image Processing
CHAPTER 1

Introduction
This book contains descriptions of algorithms for image processing such as noise
reduction, including reduction of impulse noise, contrast enhancement, shading
correction, edge detection, and many others. The source codes of the projects in the
C# programming language implementing the algorithms are included on the book’s
companion web site. The source codes are Windows Forms projects rather than
Microsoft Foundation Classes Library (MFC) projects. The controls and the graphics in
these projects are implemented by means of simple and easily understandable methods.
I have chosen this way of implementing controls and graphics services rather than those
based on MFC because the integrated development environment (IDE) using MFC is
expensive. Besides that, the software using MFC is rather complicated. It includes many
different files in a project and the user is largely unable to understand the sense and
usefulness of these files. On the contrary, Windows Forms and its utility tools are free
and are easier to understand. They supply controls and graphics similar to that of MFC.
To provide fast processing of images we transform objects of the class Bitmap, which
is standard in Windows Forms, to objects of our class CImage, the methods of which are
fast because they use direct access to the set of pixels, whereas the standard way of using
Bitmap consists of implementing the relatively slow methods of GetPixel and SetPixel
or methods using LockBits, which are fast, but not usable for indexed images.
The class CImage is rather simple: It contains the properties width, height, and nBits
of the image and methods used in the actual project. My methods are described in the
chapters devoted to projects. Here is the definition of our class CImage.

  class CImage
  { public Byte[] Grid;
     public int width, height, nBits;

     public CImage() { } // default constructor

3
© Vladimir Kovalevsky 2019
V. Kovalevsky, Modern Algorithms for Image Processing, https://doi.org/10.1007/978-1-4842-4237-7_1
Chapter 1 Introduction

     public CImage(int nx, int ny, int nbits) // constructor

     {
        width = nx;
        height = ny;
        nBits = nbits;
        Grid = new byte[width * height * (nBits / 8)];
    }

    public CImage(int nx, int ny, int nbits, byte[] img) // constructor
    {
        width = nx;
        height = ny;
        nBits = nbits;
        Grid = new byte[width * height * (nBits / 8)];
        for (int i = 0; i < width * height * nBits / 8; i++) Grid[i] = img[i];
    }
  } //*********************** end of class CImage *****************

Methods of the class CImage are described in the descriptions of projects.

4
CHAPTER 2

Noise Reduction
Digital images are often distorted by random errors usually referred to as noise. There
are two primary kinds of noise: Gaussian noise and impulse noise (see Figure 2-1).
Gaussian noise is statistical noise having a probability distribution similar to a Gaussian
distribution. It arises mainly during acquisition (e.g., in a sensor). It could be caused
by poor illumination or by high temperature of the sensor. It comes from many natural
sources, such as the thermal vibrations of atoms in conductors, referred to as thermal
noise. It influences all pixels of the image.
Impulse noise, also called salt-and-pepper noise, presents itself as sparsely occurring
light and dark pixels. It comes from pulse distortions like those coming from electrical
welding near the electronic device taking up the image or due to improper storage of old
photographs. It influences a small part of the set of pixels.

Figure 2-1. Examples of noise: (a) Gaussian noise; (b) impulse noise

5
© Vladimir Kovalevsky 2019
V. Kovalevsky, Modern Algorithms for Image Processing, https://doi.org/10.1007/978-1-4842-4237-7_2
Chapter 2 Noise Reduction

The Simplest Filter

We consider first the ways of reducing the intensity of Gaussian noise. The algorithm for
reducing the intensity of impulse noise is considered later in this chapter.
The most efficient method of reducing the intensity of Gaussian noise is replacing
the lightness of a pixel P by the average value of the lightness of a small subset of pixels
in the neighborhood of P. This method is based on the fact from the theory of random
values: The standard deviation of the average of N equally distributed random values
is by the factor √N less than the standard deviation of a single value. This method
performs a two-dimensional convolution of the image with a mask, which is an array of
weights arranged in a square of W × W pixels, and the actual pixel P lies in the middle
of the square. Source code for this filter is presented later in this chapter. This method
has two drawbacks: It is rather slow because it uses W × W additions for each pixel of the
image, and it blurs the image. It transforms fine edges at boundaries of approximately
homogeneous regions to ramps where the lightness changes linearly in a stripe whose
width is equal to W pixels. The first drawback is overcome with the fast averaging filter.
However, the second drawback is so important that it prevents use of the averaging filter
for the purpose of noise removal. Averaging filters are, however, important for improving
images with shading (i.e., those representing nonuniformly illuminated objects), as we
see later. I propose another filter for the purpose of noise removal as well.
First, let us describe the simplest averaging filter. I present the source code of this
simple method, which the reader can use in his or her program. In this code, as well
as in many other code examples, we use certain classes, which are defined in the next
section.

The Simplest Averaging Filter

The nonweighted averaging filter calculates the mean gray value in a gliding square
window of W × W pixels where W is the width and height of a square gliding window.
The greater the window size W, the stronger the suppression of Gaussian noise: The filter
decreases the noise by the factor W. The value W is usually an odd integer W = 2 × h + 1
for the sake of symmetry. The coordinates (x + xx, y + yy) of pixels inside the window vary
symmetrically around the current center pixel (x, y) in the intervals: −h ≤ xx ≤ +h and
−h ≤ yy ≤ +h with h = 1, 2, 3, 4, and so on.

6
 Chapter 2 Noise Reduction

Near the border of the image the window lies partially outside of the image. In this
case, the computation loses its natural symmetry because only pixels inside the image
can be averaged. A reasonable way to solve the border problem is to take control of the
coordinates (x + xx, y + yy) if they point out of the image. If these coordinates are out
of the image the summation of the gray values must be suspended and the divisor nS
should not be incremented.
An example of the algorithm for averaging the colors is presented here. We often use
in our code comments denoted by lines of certain symbols: lines marked with = label the
start and the end of a loop, lines with minus signs label if instructions, and so on. This
makes the structure of the code more visible.
The simplest slow version of the algorithm has four nested for loops.

public void CImage::Averaging(CImage input, int HalfWidth)

{ int nS, sum;
  for (int y=0; y<height; y++) //================================
  { for (int x=0; x<width; x++) //==============================
    { nS=sum=0;
      for (int yy=-HalfWidth; yy<=HalfWidth; yy++) //=======
      {  if (y+yy>=0 && y+yy<input.height )
          for (int xx=-HalfWidth; xx<=HalfWidth; xx++) //===
          { if (x+xx>=0 && x+xx<input.width)
            { sum+=input.Grid[x+xx+width*(y+yy)];
               nS++;
            }
          } //====== end for (xx... ===================
      } //======== end for (yy... =======================
      Grid[x+width*y]=(sum+nS/2)/nS; //+nS/2 is for rounding
    } //============ end for (x... ========================
  } //============== end for (y... ========================
} //****************** end Averaging ****************************

This is source code: The reader can copy it and put into its C# source, and it will
work.
The parameter HalfWidth is half the width of the gliding window. The width and
height of the window are both equal to 2*HalfWidth+1. The variables x and y in the
preceding code are the indexes of pixels in Grid, and xx and yy are the indexes of the
pixels inside the gliding averaging window.
7
Chapter 2 Noise Reduction

The computation of sum in the innermost for loop needs W × W additions and
W × W accesses to the image for each pixel of the input image, which is quite time
consuming.
Let us remark once again that averaging filters, although they are very efficient
at reducing the intensity of Gaussian noise, strongly blur the image. Therefore they
should not be used for noise reduction. I suggest using the sigma filter described later
in this chapter for this purpose. Averaging is used for the purpose of shading correction
(see Chapter 4). Therefore it will be mostly used with a rather large gliding window, as
large as half the width of the image. Then the simplest averaging routine becomes so
time consuming that it is practically impossible to use it. For example, in the case of a
grayscale image of 1000 × 1000 pixels and a gliding window of 400 × 400 pixels, which is
typical for shading correction, the runtime of the function Averaging on a standard PC
can take about 20 minutes.

The Fast Averaging Filter

The simplest averaging filter makes W2 additions and one division for each pixel of
the image. For example, it makes 5 · 5 + 1 = 26 operations for each pixel in the case of a
gliding window of 5 × 5 = 25 pixels. There are applications (e.g., the shading correction)
that are using much greater gliding windows; for example, one of 400 × 400 = 160,000
pixels. If an application uses the simplest averaging filter with such a great gliding
window, it can run for several minutes.
It is possible to accelerate the averaging filter using the following basic idea: It is
possible to calculate first the sums of gray values in small one-dimensional windows of
1 × W pixels. We call the arrays of these sums the columns. In the following description
of the basic idea we use a Cartesian coordinate system, the index of a column of the
image being the abscissa x and the index of a row being the ordinate y. We consider
the ordinate axis as directed downward, from top to bottom, which is usual in image
processing and different from the direction of the ordinate axis in mathematics.
The fast averaging filter solves the same problem as the simplest filter, but it is much
faster. The fast filter blurs the image in the same way as the simplest filter does. Therefore
its main application is shading correction rather than noise suppression.
When using the basic idea just mentioned, it is possible to reduce the number of
operations per pixel from W × W to ≈4: The fast filter calculates and saves the sum of the
gray values in each column of W pixels, and the middle pixel of each column lies in the

8
 Chapter 2 Noise Reduction

actual row of the image (Figure 2-2). The filter then directly calculates the sum over the
window having its central pixel at the beginning of a row; that is, by adding up the sums
saved in the columns. Then the window moves one pixel along the row, and the filter
calculates the sum for the next location by adding the value of the column sum at the
right border of the window and by subtracting the value of the column sum at the left
border. It is necessary to check whether the column to be added or subtracted is inside
the image. If it is not, the corresponding addition or subtraction must be skipped.

Old
window
Old column
W×W
1×W - - - -
New
column
1×W New
window
- + W×W
Actual row

+ + + +
+Y

Figure 2-2. Explanation of the functioning of the fast average filter

Due to applying a similar procedure for the calculation of the column sums, the
average number of additions or subtractions per pixel is reduced to ≈2 + 2 = 4. The
sum inside the window must be calculated directly (i.e., by the addition of HalfWidth + 1
sums of columns) only for a pixel at the beginning of each row. The sums of columns
must be calculated directly only for the pixels of the first row of the image.
The filter updates the values of the columns when proceeding to the next row of
the image by adding the gray value below the lower end and subtracting the gray value
at the upper end of each column (Figure 2-2). In this case it is also necessary to check
whether the gray value to be added or subtracted is in the image. The filter divides (with
rounding) the sum by the number of pixels in the intersection of the window with the
image as soon as the sum of the gray values in a window is calculated and saves the
result in the corresponding pixel of the output image.

9
Visit https://ebooknice.com to
discover a wide range of
eBooks across various genres.
Enjoy exclusive deals and
discounts to enhance your
reading experience. Start your
digital reading journey today!
Chapter 2 Noise Reduction

Here is the source code of the simplest version of the fast averaging filter designed for
filtering grayscale images.

public int FastAverageM(CImage Inp, int hWind, Form1 fm1)

// Filters the grayscale image "Inp" and returns the result in 'Grid' of the
// calling image.
{
    width  = Inp.width ; height = Inp.height; // elements of class CImage
    Grid = new byte[width  * height];
    int[] SumColmn; int[] nPixColmn;
    SumColmn = new int[width];
    nPixColmn = new int[width];
    for (int i = 0; i < width ; i++) SumColmn[i] = nPixColmn[i] = 0;

    int nPixWind = 0, SumWind = 0;

    for (int y = 0; y < height + hWind; y++) //=============================
    {
      int yout = y - hWind, ysub = y - 2 * hWind - 1;
      SumWind = 0; nPixWind = 0;
      int y1 = 1 + (height + hWind) / 100;
      for (int x = 0; x < width + hWind; x++) //===========================
      {
        int xout = x - hWind, xsub = x - 2 * hWind - 1; // 1. and 2.
addition
        if (y < height && x < width )
        {
          SumColmn[x] += Inp.Grid[x + width  * y];
          nPixColmn[x]++; // 3. and 4. addition
        }
        if (ysub >= 0 && x < width )
        {
          SumColmn[x] -= Inp.Grid[x + width * ysub];
          nPixColmn[x]--;
        }
        if (yout >= 0 && x < width )
        {

10
 Chapter 2 Noise Reduction

          SumWind += SumColmn[x];
          nPixWind += nPixColmn[x];
        }
        if (yout >= 0 && xsub >= 0)
        {
          SumWind -= SumColmn[xsub];
          nPixWind -= nPixColmn[xsub];
        }
        if (xout >= 0 && yout >= 0)
          Grid[xout + width  * yout] = (byte)((SumWind + nPixWind / 2) /
nPixWind);
      } //===================== end for (int x = 0; =====================
    } //====================== end for (int y = 0; ======================
    return 1;
    } //************************* end FastAverageM ***********************

I present next the universal source code of the fast average filter designed both for
color and grayscale images. It uses the variable int nbyte, which is set to 3 for color and
to 1 for grayscale images. We define for the sum of color intensities in the gliding window
of (2*hWind + 1)2 pixels an array SumWind[3] of three elements for sums of red, green,
and blue intensities. In the case of a grayscale image, only the element SumWind[0] is
being used. We use the following variables as described next.
The location with the coordinates (c+nbyte*x, nbyte*y) is the location of a
color channel, one of red, green, or blue channels whose intensity is added to the
corresponding element of the array SumColmn. The location (c+nbyte*x, nbyte*ysub)
is that of a color channel whose intensity is to be subtracted from SumColmn. The
variable c+nbyte*x is the abscissa of the short column whose contents are to be added
to SumWind[c]. The variable c+nbyte*xsub is the abscissa of the short column whose
contents are to be subtracted from SumWind[c].

public int FastAverageUni(CImage Inp, int hWind, Form1 fm1)

// Filters the color or grayscale image "Inp" and returns the result in
// 'Grid' of calling image.
{ int c = 0, nByte = 0;
   if (Inp.N_Bits == 8) nByte = 1;
   else nByte = 3;

11
Chapter 2 Noise Reduction

   width  = Inp.width ; height = Inp.height; // elements of the class "Cimage"

   Grid = new byte[nByte * width  * height];
   int[] nPixColmn;
   nPixColmn = new int[width];
   for (int i = 0; i < width ; i++) nPixColmn[i] = 0;
   int[,] SumColmn;
   SumColmn = new int[width, 3];
   int nPixWind = 0, xout = 0, xsub = 0;
   int[] SumWind = new int[3];
   for (int y = 0; y < height + hWind; y++) //=============================
  { int yout = y - hWind, ysub = y - 2 * hWind - 1;
     nPixWind = 0;
     for (c = 0; c < nByte; c++) SumWind[c] = 0;
     int y1 = 1 + (height + hWind) / 100;
     for (int x = 0; x < width  + hWind; x++) //============================
     { xout = x - hWind;
        xsub = x - 2 * hWind - 1;    // 1. and 2. addition
        if (y < height && x < width) // 3. and 4. addition
       { for (c=0; c< nByte; c++)
            SumColmn[x, c] += Inp.Grid[c + nByte*(x + width*y)];
         nPixColmn[x]++;
      }
      if (ysub >= 0 && x < width )
      { for (c=0; c<nByte; c++)
            SumColmn[x, c] -=Inp.Grid[c+nByte*(x+ width*ysub)];
        nPixColmn[x]--;
      }
      if (yout >= 0 && x < width )
      { for (c = 0; c < nByte; c++) SumWind[c] += SumColmn[x, c];
         nPixWind += nPixColmn[x];
      }
      if (yout >= 0 && xsub >= 0)
      { for (c = 0; c < nByte; c++) SumWind[c] -= SumColmn[xsub, c];
         nPixWind -= nPixColmn[xsub];
      }

12
 Chapter 2 Noise Reduction

      if (xout >= 0 && yout >= 0)

        for (c = 0; c < nByte; c++)
          Grid[c+nByte*(xout+width*yout)]=(byte)( SumWind[c] / nPixWind);
    } //============= end for (int x = 0;  =============================
  } //============== end for (int y = 0;  ==============================
  return 1;
} //***************** end FastAverageUni ********************************

This source code can be used in a corresponding Windows Forms project. It is not
the fastest version; it can be made 50 percent faster by removing the multiplications from
the interior loop. Some multiplications can be performed before starting the loop; some
others can be replaced by additions. A still faster version can be made containing the
following nine loops instead of the two loops with the indexes y and x in FastAverageM
or in FastAverageUni:

for (int yOut=0; yOut<=hWind; yOut++)

{ for(int xOut=0; xOut<=hWind; xOut++){...} //Loop 1
   for(xOut=hWind+1; xOut<=width-hWind-1; xOut++) {...} //Loop 2
   for(xOut=width-hWind; xOut<width; xOut++) {...} //Loop 3
}
for(yOut=hWind+1; yOut<=height-hWind-1; yOut++)
{ for(xOut=0; xOut<=hWind; xOut++) {...} //Loop 4
   for(xOut=hWind+1; xOut<=width-hWind-1; xOut++) {...} //Loop 5
   for(xOut=width-hWind; xOut<width; xOut++) {...} //Loop 6
}
for(yOut=height-hWind; yOut<height; yOut++)
{ for(xOut=0; xOut<=hWind; xOut++) {...} /Loop 7
   for(xOut=hWind+1; xOut<=width-hWind-1; xOut++) {...} //Loop 8
   for(xOut=width-hWind; xOut<width; xOut++) {...} //Loop 9
}

Each of the nine loops processes a part of the image (see Figure 2-3) that is either
hWind + 1 pixels wide or hWind + 1 pixels high.

13
Chapter 2 Noise Reduction

Part 1 Part 2 Part 3

hWind+1

Part 4 Part 5 Part 6

hWind+1

Part 7 Part 8 Part 9

Figure 2-3. The nine parts of the image corresponding to the nine loops

This version of the fast averaging filter can be used only if the condition hWind ≤
min (width, height)/2 - 1 is fulfilled. In such a version of the routine the interior loops
with the variable xOut contain no multiplications and no if instructions. The routine
is about 60 percent faster than the previously described FastAverageM. However, it
is much longer and much more difficult to debug. The gain in speed is not essential:
This code uses 0.7 seconds to process a big color image of 2448 × 3264 pixels with the
gliding window of 1200 × 1200 pixels, whereas FastAverageM takes 1.16 seconds. These
calculation times are almost independent from the size of the gliding window: They are
0.68 and 1.12 seconds correspondingly for the case of a gliding window of 5 × 5 pixels.

The Fast Gaussian Filter

The averaging filter produces a smoothed image in which some rectangular shapes not
present in the original image can be seen. These shapes appear because the averaging
filter transforms each light pixel to a homogeneously light rectangle of the size of the
gliding window. As soon as a pixel has an outstanding lightness strongly differing
from the values of adjacent pixels, the rectangle becomes visible. This is an unwanted
distortion. It can be avoided when using the Gaussian filter that multiplies the gray
values to be added by values that decay with the distance from the center of the window
according to the two-dimensional Gauss law. In addition, the Gaussian filter provides a
better suppression of noise. An example of the weights is shown in Figure 2-4.

14
 Chapter 2 Noise Reduction

-Y

0.011 0.018 0.023 0.023 0.018 0.011

0.018 0.030 0.039 0.039 0.030 0.018

0.023 0.039 0.050 0.050 0.039 0.023

-X +X

0.023 0.039 0.050 0.050 0.039 0.023

0.018 0.030 0.039 0.039 0.030 0.018

0.011 0.018 0.023 0.023 0.018 0.011

+Y

Figure 2-4. Example of weights in the gliding window of the classical Gauss filter

These values are called the weights of the filter. The weights corresponding to the
two-dimensional Gauss law are floats less than one:

w(x, y) = (2πσxσy)-1exp(-x2/2σ2x - y2/2σ2y).

They can be calculated in advance and saved in a two-dimensional array whose

size corresponds to the size of the gliding window (Figure 2-4). Then the gray values
or color channels of the filtered image are multiplied with the weights and the sum of
the products is calculated. This procedure needs W2 floating point multiplications and
W2 ­additions per pixel of the grayscale image to be filtered where W is the width of the
gliding window. In the case of a color image, this number is 3W2.
There is a possibility of obtaining approximately the same results using the knowledge
of the statistics that says that the convolution of many equivalent probability distributions
tends to the Gaussian distribution. The convergence of this process is so fast that it is
sufficient to calculate the convolution of only three rectangular distributions to obtain a
good approximation. A rectangular distribution has a constant density in an interval and a
zero density outside the interval. If an image is processed with a filter three times, the result
is equivalent to the filtering with weights being convolutions of three rectangular weights.
Thus to perform approximately a Gaussian filtering of an image it is sufficient to filter the
image three times with a fast averaging filter. This procedure requires 4 × 3 = 12 integer

15
Chapter 2 Noise Reduction

additions per color channel of a pixel independent from the size of the window. We have
calculated that the standard of the equivalent Gaussian ­distribution is proportional to the
half-width of the gliding window of the averaging filter. In Table 2-1 hWind is the half-width
of the averaging window, and Sigma is the standard of a random variable whose distribution
corresponds to the weights calculated by the triple filtering with the fast filter.

Table 2-1. Relation Sigma/hWind tending to 1 with

increasing hWind

hWind 1 2 3 4 5

Sigma 1.414 2.456 3.459 4.456 5.49

Sigma/hWind 1.414 1.218 1.153 1.114 1.098

You can see that the relation Sigma/hWind tends to 1 when the width of the window
increases.
Figure 2-5 shows how the weights of the approximate Gauss filter differ from true
Gauss weights.

weights after one filtering

weights after two filtering
weights after three filtering
true Gaussian weights

Figure 2-5. Weights of the approximate and true Gauss filters

16
 Chapter 2 Noise Reduction

The Median Filter

The averaging and the Gaussian filters provide the most efficient suppression of Gaussian
noise. The averaging filter with a gliding window that has the width of W = 2h + 1 pixels
transforms steep edges of homogeneous regions to ramps of the width W. In the case of
a Gaussian filter, the ramp is steeper. However, both these filters blur the image, so they
should not be used for noise suppression.
Most textbooks on image processing recommend using median filters for noise
suppression. A median filter sorts the intensities of colors in the gliding window and
replaces the intensity in the middle of the gliding window by the intensity staying in the
middle of the sorted sequence. Median filters can also be used for suppressing impulse
noise or salt-and-pepper noise.
Almost no textbook, though, draws the attention of the reader to very important
drawbacks of the median filter. First, it heavily distorts the image. A median filter with a
gliding window of (2 * h + 1)2 pixels deletes each stripe of the width of less than h pixels.
It also deletes a triangular part of approximately 2h pixels at each corner of a rectangular
shape. Even more, it inverts a part of the image containing some parallel stripes of
the width h if the width of the spaces between the stripes is also equal to h (compare
Figure 2-6 with Figure 2-7). This is easily understandable if the reader notices that
median makes decisions according to the majority: The central pixel becomes dark if the
majority of pixels in the gliding window are dark.

Figure 2-6. Original image and the gliding window of 5 × 5 pixels

17
Chapter 2 Noise Reduction

Figure 2-7. The same image after filtering with median of 5 × 5 pixels
Using the median for the suppression of impulse noise is also not recommended
because it will delete objects having the shape of thin lines that have nothing to do with
noise. I suggest an efficient method in a later chapter.

Sigma Filter: The Most Efficient One

The sigma filter reduces noise in the same way as the averaging filter: by averaging many
gray values or colors. The idea of the sigma filter is averaging only those intensities
(i.e., gray values or intensities of color channels) in a gliding window that differ from
the intensity of the central pixel by no more than a fixed parameter called tolerance.
According to this idea, the sigma filter reduces the Gaussian noise and retains the edges
in the image not blurred.
The sigma filter was suggested by John-Sen Lee (1983). However, it remained almost
unknown until recently: It has been mentioned in no textbook for image processing that
I am aware of. It was mentioned in a professional paper only once Chochia (1984).
A filter similar to the sigma filter was suggested by Tomasi and Manduchi (1998),
which they called the bilateral filter. They suggested assigning two kinds of weights
to the colors being averaged: a domain weight becoming smaller with the increasing
distance of the averaged pixel from the central pixel of the gliding window and a range
weight becoming smaller with increasing difference between the intensities of colors
of the pixel being averaged and that of the central pixel. Both weights can be defined as
densities of the Gauss distribution. The filter works well: It reduces the Gaussian noise
and preserves the sharpness of the edges. However, it is essentially slower than the sigma
filter; for example, to process a color image of 2500 × 3500 pixels the bilateral filter needs
30 seconds, whereas the simplest sigma filter needs only 7 seconds. Thus the bilateral
filter is approximately four times slower than the sigma filter. The authors of the bilateral
filter did not mention the sigma filter among the references.
18
Exploring the Variety of Random
Documents with Different Content
 Mr. Denham is a first-cousin of Sir Keith's, and about
the same age, but not in the least like him,—very plain, shy
and brusque in manner, and rather odd in his ways. He is a
thorough "study-man," a hard reader, a hater of platforms,
and a busy organiser of Parish work,—characteristics not
always found in juxtaposition. He is rarely to be seen out of
his study, except in the cottages of the poor, in the
sickrooms of either poor or rich, and in needful Parish
gatherings. Tennis is not in his line; and his one recreation
takes the shape of long lonely walks in the country,—hardly
sufficient recreation, perhaps, in the case of so severe a
brain-toiler. He looks like a burner of midnight oil.

The poor are devoted to Mr. Denham,—and the rich are

not. Easily explained, for to the poor he is all gentleness,
and to the rich he is all shy curtness. It is a pity; since he
loses influence thereby. Yet one does not know how to
regret the hermitage-loving turn of mind which results in
such sermons as his,—no strings of platitudes strung
together in a hurry, and spun out to fill up a stereotyped
twenty minutes or half-hour, but real downright teaching,
Sunday after Sunday, in a full and systematic course. Surely
our Church means us to have such systematic teaching,
declaring to us throughout the year the "whole counsel of
God," and not merely to be fed upon stray scraps of that
"counsel," gathered up almost at random without plan or
method. But I have never had it before.

The services too are full of help and refreshment,

bright, hearty, reverent, with a good choir and
congregational singing. Everybody seems to join. There is
no lounging or lolling, and scarcely any staring about. It is
wonderful how infectious in a Church is a spirit of deep
earnestness and intense reverence.
 I often think of Sir Keith's words on any journey here,—
of the responsibility involved in greater privileges. For I do
feel that such a Church as St. John's near at hand is a real
privilege, and ought to be a real and practical help. And that
means that I ought to advance more quickly, that I ought to
become more Heavenly-minded, that I ought to live more
nearly such a life as Christ my Master lived, that I ought to
walk more fully as He walked.

Is it so, indeed? The question is a very serious one. For

if not—better far that greater privileges and means of
spiritual advance were not mine, than that having I should
fail to use them!

"Search me, O God, and know my heart; try me, and

know my thoughts, and see if there be any wicked way in
me, and lead me in the Way everlasting."

No better prayer for me than this. Of all dangers, I

dread none more than the perils of self-deception and of
spiritual stagnation.

CHAPTER XVII.
GLADYS HEPBURN'S FIRST SUCCESSES.

DIARY OF GLADYS HEPBURN.

April 23. Thursday (preceding).

 UNCLE TOM was in London on Tuesday, and he very
kindly went to see two or three publishers for me. Mr. B.
spoke a long time about the great necessity for careful
preparation and revision. And Mr. D. also spoke about that.
He said he had just refused two manuscripts, written by a
lady, for nothing but because they were so carelessly done.
If I like at any time to send him a very carefully prepared
MS., he will look at it. And Mr. F. told Uncle Tom about some
other publishers who might be the right sort.

I begin to understand that it isn't only a question about

a story being pretty good in itself, but also one has to be
careful where one sends it. Some publishers chiefly bring
out novels; and some history; and some religious tales; and
so on. And a man might refuse a book just because it didn't
suit him, while it might be the very book for another
publisher over the way.

Mr. F. also said that the better plan was to send my

MSS. straight off, instead of writing first to ask leave.
Because, naturally, if a man hasn't seen the MS., and can't
possibly judge anything about it, he is more likely to say at
once "No" than "Yes."

May 2. Saturday.—My little book is out! It costs one-

and-sixpence. Twelve copies have arrived, bound in
different colours. The first picture is really pretty, and it is
nice to see my own ideas embodied by somebody else.

So at last my wish is granted, and I am very pleased

and thankful. Only perhaps not so desperately excited in
myself as I expected to be,—not nearly so excited as others
are about it, I think. For somehow, now it has come to pass,
the thing seems quite natural. And of course I am very
much more interested in the story I am actually writing
than in one which was finished so long ago.

May 14. Thursday.—Fifteen pounds came yesterday,—

the first money I ever earned by my own brains. I have put
some into the Savings Bank, and part of it I mean to spend.
I hope I shall never get into spendthrift ways.

Several kind letters have reached me about my little

book,—and some give advice. I think I liked best of all Mr.
Wilmington saying he had nearly cried over it. Aunt Anne
Hepburn complains that the colour of her copy is very ugly,
and she points out a misprint on the fourth page. But I
don't see how one can choose out the prettiest colours for
everybody.

The little ones from Glynde House have been here this
evening, and we all had a good game of play. It was rather
fun to feel that, as I had a printed book out, nobody would
count me too childish for my age, and so I could just enjoy
myself as much as ever I liked. Was that silly?

I wonder how soon poor Maggie will hear about her MS.
She seems getting rather impatient. I don't wonder, for I
have often felt dreadfully impatient.

May 16. Saturday.—Mother and I don't think Miss

Conway looks quite so strong and bright as when she first
came. I wonder if anything is worrying her.
 It is so strange that Maggie does not grow more fond of
Miss Con. Mother and I think Miss Con delightful. And
Ramsay is growing quite absurd. At first he used to say all
sorts of hard and contemptuous things about her, as he
does about almost everybody; but now he has turned right
round, and he seems to think the ground scarcely good
enough for her to tread on. But I don't suppose Miss Con
has the least idea of his state of admiration, for he only gets
red and awkward when he sees her. If she had, how she
would laugh! She a girl of twenty-three, with the mind of a
woman of thirty, as Uncle Tom says,—and he a backward
boy of seventeen.

And yet I don't know whether she really would laugh,—

at least it would not be unkindly.

June 17. Wednesday.—Maggie's little MS. has been sent

back, as I felt sure it must be, if she wouldn't work it up
more carefully. I am very sorry, for she is so disappointed.
But the odd thing is, that she seems quite angry, too, with
the publisher. I don't understand that, because of course he
must be free to take or refuse books. And it always seems
to me that one has just to learn what one can do, by trying.
One trial doesn't settle the matter; but a good many trials
would. And if one really had not the gift, if God really had
not called one to the work,—ought one to be vexed?

Still, if I had failed instead of succeeding, I might not

find it so easy to write like this.
 July 7. Tuesday.—My book "Tom and Mary" is finished at
last,—was finished last week, I mean: and I have been
correcting hard ever since. I don't mean the MS. to have
one single untidy page. Of course that means more copying,
but it is worth while. The greatest difficulty is to think of a
good title.

It really does seem to me large enough for a five-

shilling volume: but I have not said so to anybody, for fear
of being mistaken.

We have pretty well settled what publisher to send it to.

But I don't feel very hopeful of another success so soon. It
seems more than I ought to expect. Not very good accounts
of poor Mrs. Romilly. There seems no idea of her coming
home yet. Even if she did, I should not see Nellie, for the
Romillys all go north in about a fortnight,—as soon as
Denham's holidays begin. They did talk of going sooner, but
Mr. Romilly couldn't make up his mind to it. I'm sure I don't
know what he will do when Denham has gone to school,—
only sometimes people bear a thing better when it can't be
helped than when it can be helped. Lady Denham is very
much taken with Miss Con; and Mother and I are so
pleased. Lady Denham says she is "distinguished-looking." I
believe Sir Keith admires her too, only he is so cautious and
polite that one never can know what he does truly think and
feel. I can't make out whether he cares for anybody, really,
—more than just as a pleasant acquaintance.

It provokes me, rather; and yet of course I like him,—at

least, I suppose so. He is very good, and very handsome,
and most people count him perfect. I don't think I do. And
the sort of liking that I have wouldn't make me the least
unhappy if he went away to-morrow and never came back
again. For I should have Mother still—and Nellie,—and my
dear writing,—and a great many delightful things besides.
And yet Sir Keith is a real friend of ours, and he certainly
means to be as kind as possible to everybody all round.

That is just it! I suppose I don't care to be merely one

of "everybody all round." And if I don't, it must be pride.

And yet I shouldn't wish him not to be pleasant and

polite. And I know he likes Mother,—and I like him for that.
And if he didn't always bring on such a shy fit that I can't
speak, I might perhaps think him nice to talk to.

There is one other thing that I do like in Sir Keith; and

that is, that he doesn't think himself bound to make pretty
remarks about my writings, only just to please me. As if I
were a child, wanting sugar-plums!

I don't mean that one isn't glad to have an opinion

worth having; and honestly-meant praise is pleasant. But
that is different. And it seems to me that the people whose
opinion one cares for the most are very often the most
backward of all in giving it.

July 9. Thursday.—My MS. has gone off. Oh, I do hope

and pray that it may succeed!

July 11. Saturday.—Such a kind answer has come from

the publisher, Mr. Willis, promising "immediate attention."
 July 17. Friday.—I do believe this has been the most
delightful day I ever had in my life.

First of all there came a long letter from Nellie, just like
her dear self all through.

Then at breakfast-time Mother told me that I am to

have a quarter of really good music lessons, from a master
just come to live in Glynde. It is two years since I had any.
Won't I work hard!

After breakfast, Maggie came in to ask if the children

and I would go for a long day's ramble, to the woods.
Mother said "Yes" at once, and lessons were left. We took
our lunch with us, and had all sorts of fun.

Miss Con and I were together a good deal, and I really

do begin to love her dearly. She was so sweet,—thinking
about everybody except herself. Maggie kept hanging about
Miss Millington, just exactly in the same way she used to
hang about Nellie. It provoked me then, because I wanted
more of Nellie; and it provokes me now, because I can't
endure that little Miss Millington. But anyhow it gave me
more of Miss Con.

Thyrza and I got on better than usual; only I can't help

seeing that Thyrza does not care for me; and that makes it
so difficult to be kind and bright towards her. And the twins
were as merry as could be. So we enjoyed ourselves
immensely, and we didn't get home till past six o'clock. I
haven't had such a holiday for a long while.

But then came the best of all. A letter was waiting for
me at home,—from Mr. Willis. And he offers to give me £25
for the copyright of "Tom and Mary," which he thinks will
make a 5s. book. And if I agree, it is to go to the printer's
at once.
 Oh, I am so glad and thankful! It does seem so kind of
God to answer prayer like this. I know quite well I didn't
half expect it.

CHAPTER XVIII.
SERIOUS NEWS.

DIARY OF GLADYS HEPBURN—continued.

July 27. Monday.

THE Romillys leave to-morrow, and I have seen almost

if not quite the last of them to-day. They spend one night in
London at a hotel,—a troublesome plan, we think, for such
a large party; but it allows the servants to arrive a day
sooner, and to get things ready. Beckdale House seems to
be smaller than Glynde House, and so not all the servants
will go. Rouse and Phipps are to be there of course, and two
housemaids, and what Nona calls "a local cook." The old
cook stays here in charge, with one of the housemaids,—
and the gardener and his wife will be here in their cottage
too.

I wonder how the "local cook" will answer. Mr. Romilly is

so very particular about his eating.

Maggie goes flying round to-day, forgetting everything;

and Miss Con is just as quiet as usual, forgets nothing. And
poor Mr. Romilly is in a dreadful state of fuss and fidget. He
always is before a journey. A whole mountain of luggage
went off last week, and another mountain goes to-morrow;
but still he is quite sure they won't have everything they
want, and he seems perfectly certain that nobody can be
ready in time. It is comical to hear him, for, after all, the
most likely person to be late is Mr. Romilly himself. I really
don't think he can make haste. It doesn't seem to be in
him.

I should feel their leaving much more if Nellie were

here. But she isn't; and none of the other girls can be the
same to me. I'm not sure that I don't mind most of all
saying good-bye to Miss Con. Yesterday evening, after
Church, she came in for a few minutes, and she was so very
affectionate. She said to Mother, "I shall miss you and
Gladys extremely." I know we shall miss her.

Just now I am beginning another story, and that is, of

course, a great interest.

This morning I had my first music-lesson. Mr. Lee is

rather odd; and the lesson was delicious.

He said, "Play the Scale of C in octaves." When I had

done it, he said, "Wrong, from first to last."

That made me feel rather small; because I thought I

certainly could play—well, just a little nicely. I am always
asked to play at friends' houses, and once or twice I have
even been clapped, and perhaps made to feel rather
conceited. But of course Mr. Lee is a much better judge than
the common run of people, and it must be such a good
thing to find out one's real level in anything one does. I
shall have to work hard now, to get on. And the first thing
will be to learn the "wrist-action," as he calls it, which he
thinks so much of.

His touch is just splendid. He seems to bring something

out of the piano which I never knew before to be in it. All
day long I have been hearing the ring of those wonderful
octaves and chords,—almost more exciting than the thought
of my book. It has been very hard to settle down to
anything else, and trying to write was a sham.

Wasn't it odd? I was going into the drawing-room to-

day, and I overheard Mother say—

"Gladys is very much pleased with her new music-

master."

"Placidly pleased," Uncle Tom said, and he laughed,

while Ramsay added—

"Oh, that's all one must expect. Nothing excites

Gladys."

And I turned and ran away. I felt so stupidly hurt, I

could have cried. It was stupid, for I shouldn't at all like any
one to know just exactly how I do feel, and yet one does
wish to be understood. It has made me think how very little
one person can know of another's inside, merely from his or
her outside,—and how easily I may be mistaken in others,
just as they are mistaken in me!

By-the-bye, I must be very careful not to say much

about the book before Maggie; for it might not seem kind.
She has had her MS. sent back by a second publisher. I do
wish she would take a little more trouble to do well, so as to
give herself a fair chance.
 She has an idea now of writing to some well-known
authoress, to ask for advice about getting a book published,
and for an opinion on her story. Miss Millington has put this
into Maggie's head. Miss Millington says young authors
often do it. I wonder if that is true. I never thought of trying
such a plan; and I can't fancy that it could make much
difference in the end. For, after all, one must go, sooner or
later, to publishers and editors. Still, perhaps she will get a
little advice of some sort.

July 28. Tuesday.—The Romillys are off; and I feel a

great deal more flat and dismal than I expected. Glynde
House looks so frightfully empty. I can't bear to walk past it.

We have not had a comfortable day: for Ramsay is in a

mood to rub everybody the wrong way. Because of Miss
Con, I suppose. Mother says, "Poor Ramsay!" While I am
afraid I feel more like saying, "Poor Gladys!" For when he is
like this, he makes me cross too.

Mother spoke to me this evening about giving way to

temper: and I know she is right. Another person's ill-
humours are no excuse for me. But it is very difficult. If
only people would be reasonable and sensible.

I do want not to grow horrid and conceited, just

because I have had a little success. And that, of course, is a
real danger. If I were not the least proud, I shouldn't mind
so much the things he says. And of course I ought to think
of his lameness, and of what a trial it must be to a boy not
to play at cricket and football, or to run races and do
everything like other boys. It wouldn't matter so much for a
girl, but for a boy it really is dreadful. Yet when he worries
me, I don't remember that, I only think of defending
myself.

Nellie wrote so sweetly in her last letter. She said, "You

know, darling Gladys, I am not clever, and I shouldn't like
you to think me preaching, but still I do hope that having
this work given you to do for Jesus will make you keep very
close to Him."

And oh, I do hope the same! For it is work for Jesus,—

though I am afraid some of it is for myself too, because I do
so love writing, and I do so like what it brings. But He does
give me the work to do; and I want it to be for Him; and I
want to honour Him. I must pray to be able to keep silent
when I feel vexed.

July 29. Wednesday.—Such a thing has happened,—and

I am very unhappy. And yet I am so thankful that darling
Nellie herself is not hurt.

Just before breakfast this morning a telegram came for

Mother. It was from Nellie. She and Mrs. Romilly have just
reached Cologne, where we knew they were going for a few
days on their way farther North. The telegram is from a
Cologne hotel, and it says—

"Please break news to father, railway accident, Mother

much hurt, will Eustace come?"

And that is all. Not a word about whether she is in

danger or not. But we all know that it must mean danger.
Nellie would never frighten Mr. Romilly without good reason,
—or send for Eustace.
 The telegram seems to have been delayed, for Uncle
Tom says it certainly ought to have come sooner, and that is
so unfortunate, for it might just have caught Mr. Romilly.

The question was what to do. Uncle Tom would have

gone straight up to London, but it was impossible that he
should arrive before they had all left for Yorkshire. So a
telegram was sent to Miss Con, repeating Nellie's words,
and asking her to break it. Uncle Tom also telegraphed to
somebody in the hotel, begging to hear at once whether the
other telegram had arrived in time.

But it did not. The Romillys were off first. So then Uncle
Tom sent another copy of the same telegram to the station
where we believed they would stop for lunch, and a second
copy to Beckdale Station, which is some miles off from
Beckdale House.

All day long we have been waiting for a reply. It has

been impossible to do much of anything. Of course the
children's lessons had to go on: and I took out my writing
just as usual, because I am determined not to get into the
way of being a slave to moods. But I couldn't get a single
page done, worth keeping. And every time a bell rang, one
of us ran into the passage.

We are afraid now that they will not know what has
happened, till they get to Beckdale Station. If the first
telegram had reached them, we must have heard before
this.

It is a comfort to know that Eustace has joined them in

London. But what will poor Mr. Romilly do? And to think of
Nellie, all this time alone with Mrs. Romilly among
strangers,—Mrs. Romilly perhaps dangerously ill, and only
one English maid there to be any help. It does seem very
terrible. Only I know how brave darling Nellie is, and how
she forgets herself and always seems to lean upon God,
when she is in any difficulty.

July 30. Thursday.—Quite late last night, after we had

all gone to bed, a telegram came from Miss Con. Mother
slipped into my room, to tell me, if I should be awake: and I
was. It was dated from Beckdale Station, and it only said—

"News received, Mr. R. and E. off at once to Cologne,

rest of us go to Beckdale."

I shouldn't have expected Mr. Romilly to show so much

spirit; but it seems quite right. Uncle Tom has been looking
out about trains, and he finds that Mr. Romilly and Eustace
could come south by a return-train from Beckdale Station,
not long after they got there. Most likely that is what they
have done.

CHAPTER XIX.
A MOUNTAIN STATION.

CONSTANCE CONWAY'S JOURNAL.

July 30. Thursday.

 I SHALL never forget our arrival at Beckdale Station
yesterday.

Everybody was in the highest spirits,—that is to say, the

highest of which each was capable,—charmed with the
glimpses of mountain scenery which we had during the last
hour or so of our journey. Clouds hung rather low, shutting
off the summits; but I don't know that this did not make the
views only more Impressive. For imagination was free to
add unknown altitudes, and a touch of mystery always gives
sublimity,—not alone in landscapes.

Fine rain had begun to fall when we reached Beckdale

Station, and the wind was gusty enough to send Nona's hat
bowling along the platform. Of course a chase took place,
with much laughter. The luggage was bundled out from the
van; and we found a waggonette and dog-cart in waiting,
besides two carts for the luggage.

Privately I wondered how we were all to pack into these

two vehicles, for a drive of five or six miles; but I would not
suggest difficulties, and after all the waggonette was a very
roomy one. Mr. Romilly seemed greatly disturbed at the
thought of a dog-cart for any of his party: two-wheelers
being his pet aversion: and he also showed alarm about the
steep descent before us. For the line of rail by which we had
come, since our latest change, had gradually ascended to
quite a respectable height on the mountain-sides, and a
particularly rugged road led downward from the station into
the valley,—Beckdale Station being at the head or upper
end of the long valley or dale wherein our "summer
residence" is situated.

Poor Mr. Romilly! He fidgeted up and down the platform,

counting his packages, bemoaning the deploring his choice
of this route, dolefully wondering how we should ever reach
Beckdale House. I am afraid I must confess to a sense of
amusement. Naturally I have not much sympathy with the
state of mind which insists on manufacturing troubles out of
nothing. Yet I hardly know whether any weakness is more
deserving of pity than this,—just because it is so distinctly a
character-weakness as to be seldom recognised as such by
its possessor, and therefore seldom really cured.

For two or three minutes I listened; and then I forgot all

about Mr. Romilly, standing outside the station-shelter. The
rain drove against me in fine sheets, like spray; but what
did that matter? I was revelling in my first view of
mountains. Bath hills I know well, but aught like this I have
never seen before.

Beyond the valley, on either side, there rose wild grey

heights, capped by stormy grey clouds which seemed to
drop long trains or fringes into every gorge and cleft. I think
it was the wildness, the greyness, the lonely and solemn
unworldliness of the scene, which told upon me most. The
stately march of those cloud-battalions over the mountain-
tops was grandly indescribable. I seemed to be gaining a
glimpse of something high and pure, far removed from the
littlenesses of everyday life. This small station and our tiny
selves were a mere accessory—almost a mistake.

Then all at once everyday life came back to me. For

somebody stepped up, and put a telegram-envelope into my
hand.

I thought of Albinia instantly—Albinia as ill, or perhaps

suddenly widowed. She would want me in London. Could I
go, or was I tied to my duties at Beckdale? These questions
flashed past while I opened the envelope. Elfie was close to
my side,—I had not seen her before,—and her dusky eyes
grew large with ready sympathy, as she murmured, "Poor
Miss Con! I hope it isn't anything the matter with anybody."

Anybody belonging to me, she meant. But at once I

saw.

"Elfie dear, I must speak to your father," I said quietly.

"You and Nona had better put on your waterproofs
meantime, for it is raining."

I have not the least idea how I managed to evade

inquiries, and to get into the waiting-room, alone with Mr.
Romilly and his eldest son. That came about somehow,—
Elfie assisting, under the evident impression that I had
some trouble of my own to communicate. And then I broke
the news.

Telegrams tell so cruelly little,—I have always felt this,

yet never so keenly as when I stood in the little bare
waiting-room, with the slip of paper in my hand, and those
two faces looking anxiously for more—more! A railway
accident,—of what kind we do not know; Mrs. Romilly
"much hurt,"—how much we cannot guess; Eustace wanted
there,—for what purpose we are not informed. Cologne so
far-away too. The telegram went first to the Hepburns, and
Mr. Hepburn has forwarded the sad news to me.

Eustace heard with his usual gravity; and somehow the

shock to Mr. Romilly seemed less than I had expected. He
did not lose his presence of mind, and there was not half so
much fretting over this real calamity as over the minor
worries of the journey. He said sighingly, "Poor dear
Gertrude!" And then—"But I think—er—our duty is quite
plain. Pray inquire about trains, my dear boy. I think—er—
you and I should be off with—er—as little delay as possible.
Yes, at once—er,—no need to go on to Beckdale. I could
never forgive myself, if—er—if anything happened, and I
had remained here. And Miss Con will undertake—er—the
entire management—er—"

He came to a helpless pause.

"Yes, father," assented Eustace.

"I think—er—Phipps must accompany us,—yes, certainly

—er—I cannot manage without Phipps." He sighed again
dolefully. "It is a severe strain—er—in my health. But the
call is urgent—er—undoubtedly urgent. Your dear mother is
'much hurt,' Nellie says,—and whatever that may mean—er
—my duty is, I think—plain."

The thought flashed across me, quite wickedly, that

poor little Mr. Romilly was by no means sorry to escape
"that frightful descent," as he termed the road from the
station. I cannot, of course, calmly let myself suppose that
he thought of this at all: but the idea did intrude.

"Perhaps—" he went on,—"perhaps it would be best,

Miss Conway—er,—if you could be so very kind as to call the
girls—I think it might be as well to explain—"

I obeyed with no delay, and Eustace disappeared also,

doubtless to make other arrangements. Denham was
outside, and he rushed off at my request to collect his
sisters.

"Your father has heard from abroad," I said in an

undertone. "Yes,—not quite good news; but don't frighten
any one. Only ask them all to come to the waiting-room."

I saw a general move at once in the right direction, and

went back myself to Mr. Romilly.
 Maggie entered first, rosy and laughing, her arm
through Miss Millington's,—then the twins and the little
ones, followed by Thyrza and Denham.

"My dears—er—something very sad has happened—er—

very sad indeed," Mr. Romilly began, his delicate lips
trembling like those of a distressed child. He launched into a
long and hesitating speech, which I would fain have cut
short, had it been in my power. "Your beloved mother," and
"our dear Nellie," alternated with unhappy conjectures and
dismally-expressed hopes. He read the telegram aloud, and
enlarged upon it piecemeal. Then he explained that he and
Eustace, with Phipps, would start immediately—at once—
that very evening—for London; going thence as fast as
possible to Cologne, where he hoped—er—trusted—er—to
find their precious invalid on a fair road to recovery-er.
Meanwhile—er—they were all to be dear good girls—er—and
to do exactly what their mother's friend, Miss Conway,
desired—er. He was sure he could depend upon Miss
Conway to undertake—er—all responsibility.

Eustace bent lower at this point, and said something to

him in a low voice. I had not till then noticed the return of
Eustace, and I could not hear what he said; but Mr. Romilly
nodded.

"Yes, you are right, my dear boy. It is necessary—that

one should be at the head,—in case—er—my absence
should be at all prolonged. Maggie is—er—unfortunately too
young. You understand, my dear children—all of you—" he
looked at Miss Millington among the rest,—"that during my
absence I leave—er—Miss Conway entirely responsible—and
with full authority. Yes,—with full authority—er. I wish—er—
everything to be referred to Miss Conway,—and I expect—er
—implicit obedience to her." His eyes ran over the group.
"Thyrza—you understand? If difficulties occur, the decision
will—er—will rest with Miss Conway. You understand me,
Thyrza?" He seemed to count Thyrza the only one likely to
resist my authority, whereas I knew her to be the one who
would most steadfastly uphold it.

"Yes, father," she answered. "Then I suppose Miss

Conway will have the housekeeping too."

Maggie and Miss Millington exchanged looks. Mr.

Romilly's face fell into a helpless set.

"I really—er—hardly know," he said feebly. "That is—er

—I think—er—a matter which I must leave you to—er—
settle among yourselves."

"It is unimportant—" I began; but Eustace interrupted

me—

"I beg your pardon, Miss Conway. I think Thyrza is

right. There ought to be no possibility of a mistake. While
my father is away you are, by his appointment, distinctly
and unequivocally head of the household. This includes
housekeeping. A divided authority cannot work well. For the
time, Maggie must be content to count herself one of the
girls,—subject to you. Do you not agree with me, father?"

Maggie made no protest. She only looked prettily

downcast and pensive. Mr. Romilly sighed at being appealed
to, and endorsed his son's words, though not very
emphatically. Then he went back to the telegram, and
discussed anew its meaning, with divers conjectures as to
the nature of the accident.

I do not know whether all this slow speechifying

stupefied the girls as it stupefied me. They listened for the
most part in submissive silence. Maggie's cheeks had not
lost their bloom: and though she grew serious, I am not
sure that the question of household authority did not form
the leading topic in her mind. Thyrza's face settled into a
rigid unhappiness, and Nona's eyes filled repeatedly with
tears. Elfie was the one I had feared for most; and, strange
to say, Elfie seemed the slowest of all to take alarm.
Gradually, however, a pinched misery came over her, and
the large eyes wandered about despairingly.

Eustace made use of the first pause to speak about

trains: and then I found Elfie by my side. She clutched one
of my hands, and muttered, "Can't we go? Don't let
anybody say any more."

"We will get off as soon as possible," I whispered. "Try

to be brave, Elfie. You know we may hope for brighter news
to-morrow."

"Oh, I don't know,—only don't talk,—don't let anybody

speak. Please!—please!" And she wrung her hands.

I sent her out on the platform, and had a few words

with Eustace, obtaining his help. My aim was, if I could, to
keep Elfie away from the chatterers. After some discussion,
she and I were allowed to mount the front of the dog-cart,
with only Thyrza and packages occupying the back-seat.
The remainder of our Beckdale party were packed into the
waggonette.

Once in the course of these arrangements, I found

Eustace by my side, saying something which he evidently
meant to be unheard by others.

"My father has asked me to give you these for

immediate use," he observed first: and I found bank-notes
and gold in my hand. "We will write, authorising you to
draw, if needful, on the bank for more,—if our stay should
be of any length." Then came a grateful—"It is very good of
you! We are asking you to undertake a great deal!"

"I am most glad to do all in my power," I said. "Yes, the

responsibility is heavy; I wish I had had more experience."

"It always seems that you must have had so much. But
I want to say one word. I think the girls will behave well,
and not give trouble;—still, if any difficulty should arise,
there is always Lady Denham. You could not do harm by
appealing to her. And pray write freely to Nellie. She never
makes mischief."

"Thank you very much," I said, and his warm hand-

shake was a surprise. Generally he is so undemonstrative.

It is well that these few words passed between us, for

certainly I should not have thought of Lady Denham in the
event of any difficulty. My impulse would rather have been
to appeal to Mrs. Hepburn. But evidently such an idea never
occurred to Eustace: and of course the Denhams are much
older friends.

"We shall send you news as soon as possible," Eustace

said, while tucking the wrapper well in round Elfie and me.
"Don't fret, dear;" and he kissed her cold cheek.

I was struck with his unusual freedom and almost

cheerfulness of manner. I fancy it arose from a certain
gratification in finding himself for once necessary to his
father, and useful to all of us.

"Don't fret," he repeated. "Nellie was right to send for

me: but you see, she did not mention my father, and she
would have been pretty sure to do that, if there were any
real cause for anxiety. Don't you think so?"
 Elfie tried to smile, and to say, "Yes."

CHAPTER XX.
AND A YORKSHIRE DALE.

THE SAME—continued.

OUR dog-cart took the lead, descending first the steep

and rugged road which led from the station, our driver
walking at the horse's head. I found that the waggonette
was Mr. Romilly's property, but that the dog-cart was hired,
—said driver being its owner, a kind and fatherly old farmer,
living near Beckdale House. He and his will be our nearest
neighbours here, I imagine.

Rain still fell, though not heavily; and it soon ceased,

for which I was not sorry. We could not hold up umbrellas.
The wind came in blasts.

Elfie would in a general way have been nervous to the

last degree about so abrupt a descent; but she scarcely
seemed to notice it. Her whole mind was with her mother.
The horse planted his feet slowly and with caution, his great
haunches going down in successive jerks. I could hear little
cries and exclamations of half-simulated half-real alarm
from the waggonette: yet no sound came from Elfie.
 It was getting very dusky, for we had been long at the
station. I was able, however, to see distinctly the small
sharp face leaning against me, whitey-brown in hue, with
wide-open terror-stricken eyes fixed on vacancy; and the
chill of Elfie's clasping fingers came through her gloves and
mine. I did not speak to her, however. I knew she would
rather be let alone.

The very steep hillside safely over, our drivers mounted,

and we bowled along a good road, still indeed descending,
but often so gently that one scarcely perceived the fact.

Dusky hills or perhaps mountains rose high on either

side of the long narrow dale through which our way led us:
and to our left a brawling stream rushed downward from
the dale-head. My attention was riveted, as the road
brought us near. I have never before seen a genuine
mountain-torrent. The seeing was indeed partial; but the
swirl of white foam gleamed weirdly through semi-darkness,
and the roar of small waterfalls over opposing boulders
came to my ears like grand chords of Nature's music.

I looked to discover if Elfie were able to enjoy all this;

but no, she had not stirred, and the fixed eyes were blank,
as of one whose thoughts are turned entirely inward.

Then I glanced back, to catch a glimpse of Thyrza's

straight characteristic profile—she has the best profile of
any one in the family—and she turned, as if from a
simultaneous impulse, to glance at me, her lips parted, her
eyes shining, her whole face softened into a rare new
beauty. I was so glad that enough light remained for me to
gain a clear view of her at that moment. It was a fresh
glimpse of Thyrza's real self.
 "You like this?" I breathed, leaning back to speak, and
she said—

"Oh, it is splendid!"

I do not think any more passed between us. One cannot

always talk when one feels the most.

Now and then the farmer, Mr. Stockmoor, and I

exchanged a few words across the silent Elfie. I found his
Yorkshire dialect less difficult to understand than I should
have expected; but doubtless many speak far more broadly.
He told me that there has been heavy rain lately, so the
river is unusually full. Also I learnt that the Dale is
somewhere about ten miles long; and that Beckdale House
occupies a position over six miles from the Head, and
between three and four from the farther and lower end,
where is situated the small town of Beckbergh. We might
have gone to Beckbergh Station, had it been so willed, but
an extra change of trains would have been involved thereby,
and Mr. Romilly has a mortal aversion to changes. I suspect,
however, that he will prefer to endure any number in future,
rather than commit himself to the horrors of "that frightful
descent."

The six miles seemed short to me, and I think to Thyrza

also, with such surroundings to study. July twilight,
especially in the north, is not very profound, or very quick
to deepen into darkness.

I could hardly believe that the drive was ended, when

we found ourselves entering a garden-gate and immediately
stopping before a house,—"t' hoose" our new friend called
it,—grown thickly over with creepers. It stood very close to
the road, and was a less imposing edifice than I had
perhaps expected.
 Rouse opened the door; and I explained to her at once,
briefly, how things were. Though I did not exactly state my
new position in the household, she seemed in a measure to
understand, appealing to me as well as to Maggie in respect
of sleeping arrangements. After Glynde House, the rooms
appeared limited both in number and size. Rouse had
settled matters to the best of her ability, and I thought
wisely: but Maggie at once proposed a bouleversement of
the whole. She objected to the room set apart for her use
and Thyrza's, declined to fall in with my suggestion that she
should for the present occupy the best bedroom,
complained that Miss Millington's was "horridly
uncomfortable," roamed discontentedly up and down stairs,
contradicted everybody, kept the hungry and tired children
waiting for their supper, and showed herself for once
unequivocally out of temper. Of course I knew too well the
cause; and I augured badly for the future from her mood.

It was my earnest wish to avoid needless struggles, so I

only counselled patience to the younger girls, and then did
the best I could to smooth matters by offering to share my
own bedroom with Thyrza for a few days. Maggie at first
flatly refused the offer, but gradually came round to it; and
as a next move, she requested Miss Millington to share her
bedroom—the same which she had just before denounced
as "not big enough" for two sisters.

"Instead of that poky little hole you have now, Millie

dear," she said, with a defiant glance at me from her pretty
grey eyes.

"Millie" demurred, but agreed. I made no objections,

though I could not approve of the plan. I did not believe
that Mr. or Mrs. Romilly would like her to sleep so far-away
from the children in her charge. This difficulty, however, was
removed to some extent by Thyrza's immediate—"Then I
shall sleep in that room, and leave Miss Con in peace."

She said to me later, "I know it is better for you: and of

course the little ones can't be left with no one near. Such a
shame of those two!"

I said little in answer, except to thank her for helping

me. The comfort of a room to myself still is not small,—
though indeed I would rather have Thyrza for a companion
than any one else.

To-day the girls and Miss Millington have been chiefly

occupied in unpacking and arranging. I have seen little of
any of them, for Elfie has been so ill with neuralgia, that I
could neither let her come downstairs, nor leave her alone
for any length of time.

We have had drenching rain, without a break, from

early morning. I wonder if this is usual in July. Denham has
been out, of course, regardless of soaking. Thyrza and Nona
ventured a short distance, but they were soon driven back,
and had to change everything.

Just opposite our house, on the other side of the Dale,

is a fine waterfall,—unusually full and fine just now, from
the heavy rain. It begins, not far from the top of the hill, as
a straight thread of silver. Then comes a break, where I
suppose the stream flows over a slope. Then a great
tumbling leap down a rocky descent, followed by a second
break, and by a final spreading burst of water to lower
levels, where it is hidden by trees. I could stand and watch
for hours.

The river, flowing among meadows in the bottom of the

valley, is not visible from our windows. We have glimpses of
the other road, beyond the river, running parallel with this
road; and beyond the road rises a great beautiful
mountainous mass, extending far to left and right. I do not
know its name yet, but everybody seems to call it "The
Fell." The waterfall leaps down its sides near; and far-away
to the left we can see upon it "The Scaur," a mass of
precipitous bare rock, in a green frame.

The hills on this side of the Dale, behind our house, are
more smooth and round, and less lofty. Higher up the Dale,
some miles off, we have glimpses of mountains, which I am
told are over two thousand feet in height. Their summits are
swathed in cloud at present.

July 31. Friday Afternoon.—No news yet from Germany.

I cannot make up my mind how soon we ought to hear.
Surely a second telegram might have been sent; or a letter
written by Nellie just after the accident might arrive.

Elfie was awake all night, and to-day she is shaken and
hysterical, tears springing at the least word. I would not let
her come downstairs till after lunch. Now she is on the
drawing-room sofa, sound asleep, and I am journalising at a
side-table. I feel safe in so doing, for once. We have had
another wet morning; and the sun having come out since
lunch, our whole party started ten minutes ago for a
ramble. They will not be back for at least two hours, if rain
keeps off. So I may as well utilise the time.

Thyrza alone helps me in taking care of Elfie. Maggie,

Nona, the little ones, and "Millie" keep studiously aloof. I
cannot but see that this is intentional, and that the ill-
feeling towards me is fostered by Miss Millington. Nona has
her rude manner. The little ones pout when I come near,
and are gushing towards their "sweetest darling duckie
Millie!" If I glance at Miss Millington, she bridles and tosses
her head. Maggie has scarcely spoken to me to-day, except
to oppose whatever I wish or suggest. There has been as
yet no actual resistance of my authority, and I hope that
there may not be,—also that this state of things may not
last. It is very foolish: and Maggie's ill-humour has an odd
childishness about it.

A letter unexpectedly reached me this morning from

Lady Denham,—short, but very kind. She has had a note
from Eustace, written in London; and she writes, plainly
with a clear understanding of things generally, to assure me
that I must not hesitate to appeal to her, if need should
arise.

Of course I have shown and read the letter to no one,—

though the postmark was commented on by Maggie and
Nona in what I cannot but count an unladylike because
interfering style.

If nothing else prevented me, one sentence would. Lady

Denham says, near the end—

"Is it not singular? We have just come across that

young officer again,—Captain A. Lenox, whom we saw at
Rouen. You will remember my son mentioning him in
connection with a photograph. I think you said you had
known some member of the same family. He was at the
same hotel with us in Bath, three days ago. I was glad to
see him quite well."

That is all. I do not understand her object in making the

remark. She may have written the words quite innocently,—
or she may have some dim suspicion of the truth.
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!

ebooknice.com