Fundamentals of Digital Image Processing E-Book

Chris Solomon

• Herausgeber: John Wiley & Sons
• Kategorie: Wissenschaft und neue Technologien
• Sprache: Englisch

Seitenzahl: 518

Veröffentlichungsjahr: 2011

Beschreibung

This is an introductory to intermediate level text on the science of image processing, which employs the Matlab programming language to illustrate some of the elementary, key concepts in modern image processing and pattern recognition. The approach taken is essentially practical and the book offers a framework within which the concepts can be understood by a series of well chosen examples, exercises and computer experiments, drawing on specific examples from within science, medicine and engineering.

Clearly divided into eleven distinct chapters, the book begins with a fast-start introduction to image processing  to enhance the accessibility of later topics. Subsequent chapters offer increasingly advanced discussion of topics involving more challenging concepts, with  the final chapter  looking at the application of automated image classification (with Matlab examples) .

Matlab is frequently used in the book as a tool for demonstrations, conducting experiments and for solving problems, as it is both ideally suited to this role and is widely available. Prior experience of Matlab is not required and those without access to Matlab can still benefit from the independent presentation of topics and numerous examples.

• Features a companion website www.wiley.com/go/solomon/fundamentals containing a Matlab fast-start primer, further  exercises, examples, instructor resources and accessibility to all files corresponding to the examples and exercises within the book itself.
• Includes numerous examples, graded exercises and computer experiments to support both students and instructors alike.

Leseprobe

Contents

Preface

Using the book website

1 Representation

1.1 What is an image?

1.2 Resolution and quantization

1.3 Image formats

1.4 Colour spaces

1.5 Images in Matlab

2 Formation

2.1 How is an image formed?

2.2 The mathematics of image formation

2.3 The engineering of image formation

3 Pixels

3.1 What is a pixel?

3.2 Operations upon pixels

3.3 Point-based operations on images

3.4 Pixel distributions: histograms

4 Enhancement

4.1 Why perform enhancement?

4.2 Pixel neighbourhoods

4.3 Filter kernels and the mechanics of linear filtering

4.4 Filtering for noise removal

4.5 Filtering for edge detection

4.6 Edge enhancement

5 Fourier transforms and frequency-domain processing

5.1 Frequency space: a friendly introduction

5.2 Frequency space: the fundamental idea

5.3 Calculation of the Fourier spectrum

5.4 Complex Fourier series

5.5 The 1-D Fourier transform

5.6 The inverse Fourier transform and reciprocity

5.7 The 2-D Fourier transform

5.8 Understanding the Fourier transform: frequency-space filtering

5.9 Linear systems and Fourier transforms

5.10 The convolution theorem

5.11 The optical transfer function

5.12 Digital Fourier transforms: the discrete fast Fourier transform

5.13 Sampled data: the discrete Fourier transform

5.14 The centred discrete Fourier transform

6 Image restoration

6.1 Imaging models

6.2 Nature of the point-spread function and noise

6.3 Restoration by the inverse Fourier filter

6.4 The Wiener-Helstrom Filter

6.5 Origin of the Wiener-Helstrom filter

6.6 Acceptable solutions to the imaging equation

6.7 Constrained deconvolution

6.8 Estimating an unknown point-spread function or optical transfer function

6.9 Blind deconvolution

6.10 Iterative deconvolution and the Lucy-Richardson algorithm

6.11 Matrix formulation of image restoration

6.12 The standard least-squares solution

6.13 Constrained least-squares restoration

6.14 Stochastic input distributions and Bayesian estimators

6.15 The generalized Gauss-Markov estimator

7 Geometry

7.1 The description of shape

7.2 Shape-preserving transformations

7.3 Shape transformation and homogeneous coordinates

7.4 The general 2-D affine transformation

7.5 Affine transformation in homogeneous coordinates

7.6 The Procrustes transformation

7.7 Procrustes alignment

7.8 The projective transform

7.9 Nonlinear transformations

7.10 Warping: the spatial transformation of an image

7.11 Overdetermined spatial transformations

7.12 The piecewise warp

7.13 The piecewise affine warp

7.14 Warping: forward and reverse mapping

8 Morphological processing

8.1 Introduction

8.2 Binary images: foreground, background and connectedness

8.3 Structuring elements and neighbourhoods

8.4 Dilation and erosion

8.5 Dilation, erosion and structuring elements within Matlab

8.6 Structuring element decomposition and Matlab

8.7 Effects and uses of erosion and dilation

8.8 Morpholoqical openinq and closinq

8.9 Boundary extraction

8.10 Extracting connected components

8.11 Region filling

8.12 The hit-or-miss transformation

8.13 Relaxing constraints in hit-or-miss: ‘don’t care’ pixels

8.14 Skeletonization

8.15 Opening by reconstruction

8.16 Grey-scale erosion and dilation

8.17 Grey-scale structuring elements: general case

8.18 Grey-scale erosion and dilation with flat structuring elements

8.19 Grey-scale opening and closing

8.20 The top-hat transformation

8.21 Summary Exercises

9 Features

9.1 Landmarks and shape vectors

9.2 Single-parameter shape descriptors

9.3 Signatures and the radial Fourier expansion

9.4 Statistical moments as region descriptors

9.5 Texture features based on statistical measures

9.6 Principal component analysis

9.7 Principal component analysis: an illustrative example

9.8 Theory of principal component analysis: version 1

9.9 Theory of principal component analysis: version 2

9.10 Principal axes and principal components

9.11 Summary of properties of principal component analysis

9.12 Dimensionality reduction: the purpose of principal component analysis

9.13 Principal components analysis on an ensemble of digital images

9.14 Representation of out-of-sample examples using principal component analysis

9.15 Key example: eigenfaces and the human face

10 Image Segmentation

10.1 Image segmentation

10.2 Use of image properties and features in segmentation

10.3 Intensity thresholding

10.4 Region growing and region splitting

10.5 Split-and-merge algorithm

10.6 The challenge of edge detection

10.7 The Laplacian of Gaussian and difference of Gaussians filters

10.8 The Canny edge detector

10.9 Interest operators

10.10 Watershed segmentation

10.11 Segmentation functions

10.12 Image segmentation with Markov random fields

11 Classification

11.1 The purpose of automated classification

11.2 Supervised and unsupervised classification

11.3 Classification: a simple example

11.4 Design of classification systems

11.5 Simple classifiers: prototypes and minimum distance criteria

11.6 Linear discriminant functions

11.7 Linear discriminant functions in N dimensions

11.8 Extension of the minimum distance classifier and the Mahalanobis distance

11.9 Bayesian classification: definitions

11.10 The Bayes decision rule

11.11 The multivariate normal density

11.12 Bayesian classifiers for multivariate normal distributions

11.13 Ensemble classifiers

11.14 Unsupervised learning: k-means clustering

Further reading

Index

This edition first published 2011, © 2011 by John Wiley & Sons, Ltd

Wiley-Blackwell is an imprint of John Wiley & Sons, formed by the merger of Wiley’s global Scientific, Technical and Medical business with Blackwell Publishing.

Registered office: John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK

Editorial Offices:9600 Garsington Road, Oxford, OX4 2DQ, UK 111 River Street, Hoboken, NJ 07030-5774, USA

For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com/wiley-blackwell

The right of the author to be identified as the author of this work has been asserted in accordance with the Copyright, Designs and Patents Act 1988.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher.

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books.

Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought.

MATLAB® is a trademark of The MathWorks, Inc. and is used with permission. The MathWorks does not warrant the accuracy of the text or exercises in this book. This book’s use or discussion of MATLAB® software or related products does not constitute endorsement or sponsorship by The MathWorks of a particular pedagogical approach or particular use of the MATLAB® software.

Library of Congress Cataloguing-in-Publication Data

Solomon, Chris and Breckon, TobyFundamentals of digital image processin: a practical approach with examples in Matlab / Chris Solomon and Toby Breckonp. cm.Includes index.

Summary: “Fundamentals of Digital Image Processing is an introductory text on the science of image processing and employs the Matlab programming language to illustrate some of the elementary, key concepts in modern image processing and pattern recognition drawing on specific examples from within science, medicine and electronics”— Provided by publisher.

ISBN 978-0-470-84472-4 (hardback) – ISBN 978-0-470-84473-1 (pbk.)1. Image processing–Digital techniques. 2. Matlab. I. Breckon, Toby. II. Title.TA1637.S65154 2010621.36’7—dc22

2010025730

Preface

Scope of this book

This is an introductory text on the science (and art) of image processing. The book also employs the Matlab programming language and toolboxes to illuminate and consolidate some of the elementary but key concepts in modern image processing and pattern recognition.

The authors are firm believers in the old adage, “Hearand forget...,Seeand remember..Doand know”. For most of us, it is through good examples and gently guided experimentation that we really learn. Accordingly, the book has a large number of carefully chosen examples, graded exercises and computer experiments designed to help the reader get a real grasp of the material. All the program code (.m files) used in the book, corresponding to the examples and exercises, are made available to the reader/course instructor and may be downloaded from the book’s dedicated web site – www.fundipbook.com.

Who is this book for?

For undergraduate and graduate students in the technical disciplines, for technical professionals seeking a direct introduction to the field of image processing and for instructors looking to provide a hands-on, structured course. This book intentionally starts with simple material but we also hope that relative experts will nonetheless find some interesting and useful material in the latter parts.

Aims

What then are the specific aims of this book ? Two of the principal aims are –

To introduce the reader to some of the key concepts and techniques of modern image processing.To provide a framework within which these concepts and techniques can be understood by a series of examples, exercises and computer experiments.

These are, perhaps, aims which one might reasonably expect from any book on a technical subject. However, we have one further aim namely to provide the reader with the fastest, most direct route to acquiring a real hands-on understanding of image processing. We hope this book will give you a real fast-start in the field.

Assumptions

We make no assumptions about the reader’s mathematical background beyond that expected at the undergraduate level in the technical sciences – ie reasonable competence in calculus, matrix algebra and basic statistics.

Why write this book?

There are already a number of excellent and comprehensive texts on image processing and pattern recognition and we refer the interested reader to a number in the appendices of this book. There are also some exhaustive and well-written books on the Matlab language. What the authors felt was lacking was an image processing book which combines a simple exposition of principles with a means to quickly test, verify and experiment with them in an instructive and interactive way.

In our experience, formed over a number of years, Matlab and the associated image processing toolbox are extremely well-suited to help achieve this aim. It is simple but powerful and its key feature in this context is that it enables one to concentrate on the image processing concepts and techniques (i.e. the real business at hand) while keeping concerns about programming syntax and data management to a minimum.

What is Matlab?

Matlab is a programming language with an associated set of specialist software toolboxes. It is an industry standard in scientific computing and used worldwide in the scientific, technical, industrial and educational sectors. Matlab is a commercial product and information on licences and their cost can be obtained direct by enquiry at the web-site www.mathworks.com. Many Universities all over the world provide site licenses for their students.

What knowledge of Matlab is required for this book?

Matlab is very much part of this book and we use it extensively to demonstrate how certain processing tasks and approaches can be quickly implemented and tried out in practice. Throughout the book, we offer comments on the Matlab language and the best way to achieve certain image processing tasks in that language. Thus the learning of concepts in image processing and their implementation within Matlab go hand-in-hand in this text.

Is the book any use then if I don’t know Matlab?

Yes. This is fundamentally a book about image processing which aims to make the subject accessible and practical. It is not a book about the Matlab programming language. Although some prior knowledge of Matlab is an advantage and will make the practical implementation easier, we have endeavoured to maintain a self-contained discussion of the concepts which will stand up apart from the computer-based material.

If you have not encountered Matlab before and you wish to get the maximum from this book, please refer to the Matlab and Image Processing primer on the book website (http://www.fundipbook.com). This aims to give you the essentials on Matlab with a strong emphasis on the basic properties and manipulation of images.

Thus, you do not have to be knowledgeable in Matlab to profit from this book.

Practical issues

To carry out the vast majority of the examples and exercises in the book, the reader will need access to a current licence for Matlab and the Image Processing Toolbox only.

Features of this book and future support

This book is accompanied by a dedicated website (http://www.fundipbook.com). The site is intended to act as a point of contact with the authors, as a repository for the code examples (Matlab .m files) used in the book and to host additional supporting materials for the reader and instructor.

About the authors

Chris Solomon gained a B.Sc in theoretical physics from Durham University and a Ph.D in Medical imaging from the Royal Marsden Hospital, University of London. Since 1994, he has been on the Faculty at the School of Physical Sciences where he is currently a Reader in Forensic Imaging. He has broad research interests focussing on evolutionary and genetic algorithms, image processing and statistical learning methods with a special interest in the human face. Chris is also Technical Director of Visionmetric Ltd, a company he founded in 1999 and which is now the UK’s leading provider of facial composite software and training in facial identification to police forces. He has received a number of UK and European awards for technology innovation and commercialisation of academic research.

Toby Breckon holds a Ph.D in Informatics and B.Sc in Artificial Intelligence and Computer Science from the University of Edinburgh. Since 2006 he has been a lecturer in image processing and computer vision in the School of Engineering at Cranfield University. His key research interests in this domain relate to 3D sensing, real-time vision, sensor fusion, visual surveillance and robotic deployment. He is additionally a visiting member of faculty at Ecole Supérieure des Technologies Industrielles Avancées (France) and has held visiting faculty positions in China and Japan. In 2008 he led the development of image-based automatic threat detection for the winning Stellar Team system in the UK MoD Grand Challenge. He is a Chartered Engineer (CEng) and an Accredited Imaging Scientist (AIS) as an Associate of the Royal Photographic Society (ARPS).

Thanks

The authors would like to thank the following people and organisations for their various support and assistance in the production of this book: the authors families and friends for their support and (frequent) understanding, Professor Chris Dainty (National University of Ireland), Dr. Stuart Gibson (University of Kent), Dr. Timothy Lukins (University of Edinburgh), The University of Kent, Cranfield University, VisionMetric Ltd and Wiley- Blackwell Publishers.

For further examples and exercises see http://www.fundipbook.com

Using the book website

There is an associated website which forms a vital supplement to this text. It is:

www.fundipbook.com

The material on the site is mostly organised by chapter number and this contains –

EXERCISES: intended to consolidate and highlight concepts discussed in the text. Some of these exercises are numerical/conceptual, others are based on Matlab.

SUPPLEMENTARY MATERIAL: Proofs, derivations and other supplementary material referred to in the text are available from this section and are intended to consolidate, highlight and extend concepts discussed in the text.

Matlab CODE: The Matlab code to all the examples in the book as well as the code used to create many of the figures are available in the Matlab code section.

IMAGE DATABASE: The Matlab software allows direct access and use to a number of images as an integral part of the software. Many of these are used in the examples presented in the text.

We also offer a modest repository of images captured and compiled by the authors which the reader may freely download and work with. Please note that some of the example Matlab code contained on the website and presented in the text makes use of these images. You will therefore need to download these images to run some of the Matlab code shown.

We strongly encourage you to make use of the website and the materials on it. It is a vital link to making your exploration of the subject both practical and more in-depth. Used properly, it will help you to get much more from this book.