This action might not be possible to undo. Are you sure you want to continue?
Pratt Copyright © 2001 John Wiley & Sons, Inc. ISBNs: 0-471-37407-5 (Hardback); 0-471-22132-5 (Electronic)
DIGITAL IMAGE PROCESSING
New York • Chichester • Weinheim • Brisbane • Singapore • Toronto . Inc. California A Wiley-Interscience Publication JOHN WILEY & SONS. PRATT PixelSoft.DIGITAL IMAGE PROCESSING PIKS Inside Third Edition WILLIAM K. INC. Los Altos.
. All rights reserved. NY 10158-0012. Requests to the Publisher for permission should be addressed to the Permissions Department. . ISBN 0-471-22132-5 This title is also available in print as ISBN 0-471-37407-5.com. stored in a retrieval system or transmitted in any form or by any means. It is sold with the understanding that the publisher is not engaged in rendering professional services. No part of this publication may be reproduced. except as permitted under Sections 107 or 108 of the 1976 United States Copyright Act. however. the services of a competent professional person should be sought.. New York. decompiling.Designations used by companies to distinguish their products are often claimed as trademarks. the product names appear in initial capital or all capital letters. In all instances where John Wiley & Sons. 605 Third Avenue. For more information about Wiley products. should contact the appropriate companies for more complete information regarding trademarks and registration. without the prior written permission of the Publisher. Inc. printing. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. John Wiley & Sons. including uploading. Inc.COM. E-Mail: PERMREQ @ WILEY. recording or otherwise.. is aware of a claim.Wiley. (212) 850-6011. Readers. If professional advice or other expert assistance is required. Copyright © 2001 by John Wiley and Sons. Inc. downloading. New York. visit our web site at www. fax (212) 850-6008. electronic or mechanical.
Shelly whose image needs no enhancement .To my wife.
26 Visual Phenomena. 23 Eye Physiology. 10 Image Stochastic Characterization. 29 Monochrome Vision Model. 49 45 vii .1 Photometry.2 Color Matching.1 1.1 2.CONTENTS Preface Acknowledgments PART 1 CONTINUOUS IMAGE CHARACTERIZATION 1 Continuous Image Mathematical Characterization 1. 45 3. 15 xiii xvii 1 3 Psychophysical Vision Properties 2.4 2 Image Representation.2 1. 33 Color Vision Model. 39 23 3 Photometry and Colorimetry 3.3 1. 5 Two-Dimensional Fourier Transform.5 Light Perception.3 2. 3 Two-Dimensional Systems.4 2.2 2.
147 6.1 Image Sampling and Reconstruction Concepts. 200 Karhunen–Loeve Transform.2 Image Sampling Systems.1 5. 110 5 Discrete Image Mathematical Representation 5.4 Tristimulus Value Transformation. 177 Superposition and Convolution Operator Relationships.2 5.4 5. 123 Image Statistical Characterization. 99 4. 161 Sampled Image Superposition and Convolution.3 Colorimetry Concepts. 54 3.1 Transform Domain Processing.2 7.5 Vector-Space Image Representation.3 8. 170 Circulant Superposition and Convolution. 132 Linear Operator Statistical Representation.3 Image Reconstruction Systems.4 Finite-Area Superposition and Convolution.5 General Unitary Transforms. 63 PART 2 4 DIGITAL IMAGE CHARACTERIZATION 89 91 Image Sampling and Reconstruction 4.3 7. 127 Image Probability Density Models. 91 4.5 Color Spaces. 189 Cosine.1 8.2 Transform Domain Superposition. and Hartley Transforms.2 Processing Quantized Variables. 136 121 6 Image Quantization 6. 150 141 PART 3 7 DISCRETE TWO-DIMENSIONAL LINEAR PROCESSING 159 161 Superposition and Convolution 7. 213 9.2 8. and Daubechies Transforms.3 5. Haar. 61 3. 180 8 Unitary Transforms 8.1 Scalar Quantization.viii CONTENTS 3. 207 185 9 Linear Processing Techniques 9.3 Monochrome and Color Image Quantization. Sine. 216 213 . 121 Generalized Two-Dimensional Linear Operator.1 7. 185 Fourier Transform. 195 Hadamard. 141 6.4 8.
1 12. 389 Geometrical Image Resampling. and Rotation. Minification. 221 9. 404 14.3 10.1 Binary Image Connectivity.4 General Image Restoration Models.6 Contrast Manipulation.3 Fast Fourier Transform Convolution. 236 PART 4 10 IMAGE IMPROVEMENT 241 243 Image Enhancement 10.4 Fourier Transform Filtering.7 Sensor and Display Point Nonlinearity Correction.CONTENTS ix 9. 401 14.1 13. 278 Color Image Enhancement. 304 Discrete Image Restoration Models. 300 Photographic Process Models.2 12. 261 Edge Crispening. 253 Noise Cleaning.1 11.4 13. 355 Constrained Image Restoration.3 11.1 10. 349 Statistical Estimation Spatial Image Restoration. 229 9.2 10.4 10. and Thickening. 411 .5 Translation. 284 Multispectral Image Enhancement. 297 Optical Systems Models.2 13. Skeletonizing.2 Binary Image Hit or Miss Transformations. 371 Spatial Warping. 335 SVD Pseudoinverse Spatial Image Restoration. 358 Blind Image Restoration.5 10.6 12. 363 319 13 Geometrical Image Modification 13. 386 Camera Imaging Model.3 12. 312 297 12 Point and Spatial Image Restoration Techniques 12.3 13. 289 11 Image Restoration Models 11.2 11. Magnification. 393 IMAGE ANALYSIS 371 PART 5 14 399 401 Morphological Image Processing 14.5 12. 325 Pseudoinverse Spatial Image Restoration. 243 Histogram Modification.3 Binary Image Shrinking.4 12. Thinning. 319 Continuous Image Spatial Filtering Restoration. 382 Perspective Transformation.5 Small Generating Kernel Convolution.
4 15.3 18. 485 Color Edge Detection.4 Template Matching. 443 First-Order Derivative Edge Detection.5 Binary Image Close and Open Operations. 519 Visual Texture Discrimination.3 16.3 19.1 15.2 16. 516 Texture Definition. and Area Measurements.1 16.1 18.1 19. 580 Segment Labeling. 499 Line and Spot Detection. 591 Spatial Moments. 625 613 . 616 Matched Filtering of Discrete Images.2 18. 521 Texture Features. 433 14.6 Image Feature Evaluation. 552 Clustering Segmentation Methods.7 16 Edge.2 17.x CONTENTS 14.2 19.6 15.6 Gray Scale Image Morphological Operations. 511 Transform Coefficient Features. 560 Region Segmentation Methods. 422 14. 589 Distance.2 15. 435 15 Edge Detection 15.5 Topological Attributes.3 17.5 15.4 Binary Image Generalized Dilation and Erosion. 581 551 18 Shape Analysis 18. 529 17 Image Segmentation 17.4 17. Line.3 15. 448 Second-Order Derivative Edge Detection.4 18. 607 Fourier Descriptors. 482 Luminance Edge Detector Performance.1 17.5 16. Perimeter. 609 589 19 Image Detection and Registration 19. 562 Boundary Detection. and Spot Models. 509 Amplitude Features.6 Amplitude Segmentation Methods. 499 509 443 Image Feature Extraction 16.4 16. 623 Image Registration. 469 Edge-Fitting Edge Detection.5 17. 613 Matched Filtering of Continuous Images. 597 Shape Orientation Descriptors. 566 Texture Segmentation.
685 21.11 Image Restoration Models Exercises. 691 21. 676 21. 687 21.2 PIKS Core Overview.12 Image Restoration Exercises. 674 21. 682 21.CONTENTS xi PART 6 20 IMAGE PROCESSING SOFTWARE 641 643 PIKS Image Processing Software 20.13 Geometrical Image Modification Exercises. 687 21. 692 673 Appendix 1 Appendix 2 Appendix 3 Bibliography Index Vector-Space Algebra Concepts Color Coordinate Conversion Image Error Measures 693 709 715 717 723 . 689 21. 675 21.17 Image Segmentation Exercises.15 Edge Detection Exercises.5 Image Measurement Exercises. 681 21.8 Unitary Transform Exercises.1 PIKS Functional Overview. 686 21. 663 21 PIKS Image Processing Programming Exercises 21. 678 21.3 Colour Space Exercises. 690 21. 682 21.19 Image Detection and Registration Exercises.6 Quantization Exercises. 680 21.4 Region-of-Interest Exercises. 679 21.16 Image Feature Extration Exercises.2 Image Manipulation Exercises. 643 20. 683 21.18 Shape Analysis Exercises.10 Image Enhancement Exercises.7 Convolution Exercises.1 Program Generation Exercises. 691 21.14 Morphological Image Processing Exercises.9 Linear Processing Exercises.
many image processing techniques were of academic interest only. Today. thanks to algorithmic and implementation advances. image processing has become a mature engineering discipline. in this beginning of the twenty-first century. their execution was too slow and too costly. Finally. At the beginning of that decade. and government applications. The 1980s have been a decade of significant growth and maturity in this field. I find the quoted statement still to be valid. delete content of marginal interest. Now. These exercises can be implemented using the Programmer’s Imaging Kernel System (PIKS) application program interface (API). industrial. this third edition includes computer programming exercises to bolster its theoretical content. Another motivating factor is the inclusion of interactive. But advances in the theoretical basis of image processing continue. important topics. and add discussion of new.PREFACE In January 1978. computer display imaging examples to illustrate image processing concepts. speed. in the preface to the second edition. In January 1991. Image processing has found a significant role in scientific. and cost effectiveness of digital computers and related signal processing technologies. I stated: Thirteen years later as I write this preface to the second edition. Some of the reasons for this third edition of the book are to correct defects in the second edition. image processing has become a vital cost-effective technology in a host of applications. space. PIKS is an International xiii . I began the preface to the first edition of Digital Image Processing with the following statement: The field of image processing has grown considerably during the past decade with the increased utilization of imagery in myriad applications coupled with improvements in the size.
new topics include an expanded description of color spaces. Image processing programming exercises are included in Part 6. and photometry and colorimetry. The PIKS Core version is included on a CD affixed to the back cover of this book. watershed and snake image segmentation. Hadamard. Many of the photographic examples in the book are supplemented by executable programs for which readers can adjust algorithm parameters and even substitute their own source images. Also. and practicing engineers and scientists who use image processing as a tool in their applications. Proficiency in C language programming is necessary for execution of the image processing programming exercises using PIKS. Mathematical derivations are provided for most algorithms. the Hartley and Daubechies transforms. Part 1 contains three chapters concerned with the characterization of continuous images. Among the most prominent omissions are the topics of pattern recognition. point and spatial restoration. The final chapter in Part 3 analyzes and compares linear processing techniques implemented by direct convolution and Fourier domain filtering. In addition to Part 6. image sampling and quantization techniques are explored along with the mathematical representation of discrete images. Part 4 presents a discussion of image enhancement and restoration techniques.xiv PREFACE Standards Organization (ISO) standard library of image processing operators and associated utilities. the psychophysical properties of human vision. The next two parts of the book cover the two principal application areas of image processing.” concentrates on the extraction of information from an image. and Mellin transform matched filtering. developers of image processing hardware and software systems. This third edition represents a major revision of the second edition. vector space algebra. This part describes the PIKS API and explains its use as a means of implementing image processing algorithms. Part 5. Specific topics include morphological image processing. The book is intended to be an “industrial strength” introduction to digital image processing to be used as a text for an electrical engineering or computer science course in the subject. wavelet filtering. Topics include the mathematical representation of continuous images. object shape analysis. In Part 2. including general linear operators and unitary transforms such as the Fourier. it can be used as a reference manual for scientists who are engaged in image processing research. image reconstruction from projections. and Karhunen–Loeve transforms. The reader is assumed to have a basic background in linear system theory. and random processes. The book is divided into six parts. and geometrical image modification. Although readers should find this book reasonably comprehensive. . edge detection. Part 3 discusses two-dimensional signal processing techniques. The first three parts cover the basic technologies that are needed to support image processing applications. Part 6 discusses the software implementation of image processing applications. including restoration models. image understanding. and object detection. entitled “Image Analysis. image feature extraction. image segmentation. many important topics allied to the field of digital image processing have been omitted to limit the size and cost of the book.
PRATT Los Altos. WILLIAM K. References to some of these topics are provided in the bibliography. California August 2000 . and computer graphics. scientific visualization.PREFACE xv image coding.
Laws. Davisson. John Roese. Andrews. Image processing research at USC began in 1962 on a very modest scale. Wen-Hsiung Chen. but the program increased in size and scope with the attendant international interest in the field. Keith Price.” I wish to thank the many students who suffered through the xvii . Anil K. Richard P. Michael N. Kruger. Olivier Faugeras. Dr. Ali Habibi. I wish to thank the following past and present members of the Institute’s scientific staff who rendered invaluable assistance in the preparation of the firstedition manuscript: Jean-François Abramatic. In 1971. and Robert H. Ramakant Nevatia. Zohrab Kaprielian. Nasser E. Harry C. I sincerely acknowledge the technical help of my graduate students at USC during preparation of the first edition: Ikram Abdou. then dean of engineering and vice president of academic research and administration. Nahi. Behnam Ashjari. Werner Frei. Kenneth I. Wallis. Huhns. The first edition was the outgrowth of notes developed for the USC course “Image Processing. Also. Clanton Mancill. In addition. Faramarz Davarian. Sawchuk. Lee D. announced the establishment of the USC Image Processing Institute.ACKNOWLEDGMENTS The first edition of this book was written while I was a professor of electrical engineering at the University of Southern California (USC). Nelson Mascarenhas. and Lloyd R. This environment contributed significantly to the preparation of the first edition. Jain. Sang Uk Lee. Robinson. Clifford Reader. I am deeply grateful to Professor Kaprielian for his role in providing university support of image processing and for his personal interest in my career. Welsh. Alexander A. Guner S.
The early contributors to its hardware and software design were William Bryant. Contributors to its development include Timothy Butler. to commercialize the PIKS standard. For this support. In 1988. France and at the Université de Paris. which provided partial financial support of my research at USC. Much of the PIKS content is present in this book. I believe. Andrews. Hubble. In 1996. I collaborated with Stephen A.. and Gerard A.. I joined Photon Dynamics. California. I left USC in 1979 with the intention of forming a company that would put some of my research ideas into practice. I performed sabbatical research at the Institut de Recherche d’Informatique et Automatique in LeChesney. a manufacturer of machine vision equipment for the inspection of electronics displays and printed circuit boards. Work on the second edition began in 1986. Stephen K. of Los Altos. The PIKS Core version of the PixelSoft implementation is affixed to the back cover of this edition. Hubble. Larry R. Compression Labs. Paquette. Many of the photographic examples in this book were processed on a VICOM. and Gerard A. There I worked on the development of facsimile and video coding products with Dr. . With regard to the first edition. I formed PixelSoft. Howell and Ihtisham Kabir on the development of image processing software. Paquette. California.. My research was partially supported by these institutions. and Gerard A. I formed Vicom Systems. I wish to offer words of appreciation to the Information Processing Techniques Office of the Advanced Research Projects Agency. Inc. Inc. Howell. who taught the course. During my time at Sun. has significantly enriched this third edition. Robert H. Patrick Krolak. I joined a startup company. and Ernest L. Howard Halverson. of Mountain View. There. and a Guggenheim Foundation fellowship. for their assistance in this industrial phase of my career. Inc. and many others too numerous to mention. California. to manufacture and market the VICOM image processor. In 1993. of San Jose. Sunil S.xviii ACKNOWLEDGMENTS early versions of the notes for their valuable comments. I wish to thank all those previously cited. directed by Larry G. of San Jose.. of San Jose. and William Zech. I collaborated with Larry R. Roberts. I joined Sun Microsystems. Toward that end. California. I directed a design team that developed a digital image processor called VICOM. Jeffrey Shaw. Wen-Hsiung Chen and Dr. Wallis. Paquette on the development of several hardware and software products based on PIKS. Adrian Clark. California. At Sun. Sawkar. I participated in the specification of the Programmers Imaging Kernel application program interface which was made an International Standards Organization standard in 1994. During the academic year 1977–1978. Werner Frei. In 1981. This industrial experience. Hall. Concurrently. USC. Some of the principal contributors to PIKS include Timothy Butler. Inc. Also. Having participated in the design of hardware and software products has been an arduous but intellectually rewarding task.. I appreciate the reviews of the notes provided by Harry C. I am indebted. Ali Habibi.
for her support in the writing of the third edition. Most of all. thanks are given to readers of the first two editions who reported errors both typographical and mental. I wish to thank my wife. and to Kris Pendelton. P. K.ACKNOWLEDGMENTS xix I offer my appreciation to Ray Schmidt. Also. Shelly. who created much of the line art. . W. who was responsible for many photographic reproductions in the book.