Digital image processing 4th global edition by gonzales

Digital image processing 4th global edition by gonzales Digital image processing 4th global edition by gonzales Digital image processing 4th global edition by gonzales Digital image processing 4th global edition by gonzales Digital image processing 4th global edition by gonzales

GLOBAL EDITION Digital Image Processing For these Global Editions, the editorial team at Pearson has collaborated with educators across the world to address a wide range of subjects and requirements, equipping students with the best possible learning tools This Global Edition preserves the cutting-edge approach and pedagogy of the original, but also features alterations, customization, and adaptation from the North American version GLOBAL EDITION Digital Image Processing FOURTH EDITION Rafael C Gonzalez • Richard E Woods FOURTH EDITION Gonzalez Woods GLOBAL EDITION This is a special edition of an established title widely used by colleges and universities throughout the world Pearson published this exclusive edition for the benefit of students outside the United States and Canada If you purchased this book within the United States or Canada, you should be aware that it has been imported without the approval of the Publisher or Author The Global Edition is not supported in the United States and Canada Pearson Global Edition Gonzalez_04_1292223049_Final.indd 11/08/17 5:27 PM Support Package for Digital Image Processing Your new textbook provides access to support packages that may include reviews in areas like probability and vectors, tutorials on topics relevant to the material in the book, an image database, and more Refer to the Preface in the textbook for a detailed list of resources Follow the instructions below to register for the Companion Website for Rafael C Gonzalez and Richard E Woods’ Digital Image Processing, Fourth Edition, Global Edition 1.  Go to www.ImageProcessingPlace.com 2.  Find the title of your textbook 3. Click Support Materials and follow the on-screen instructions to create a login name and password Use the login name and password you created during registration to start using the digital resources that accompany your textbook IMPORTANT: This serial code can only be used once This subscription is not transferrable Gonzalez_04_1292223049_ifc_Final.indd 11/08/17 5:33 PM D igital Image Processing Global Edition FOURTH EDITION Rafael C Gonzalez University of Tennessee Richard E Woods Interapptics 330 Hudson Street, New York, NY 10013 DIP4E_GLOBAL_Print_Ready.indb 7/6/2017 10:55:08 AM Senior Vice President Courseware Portfolio Management: Marcia J Horton Director, Portfolio Management: Engineering, Computer Science & Global Editions: Julian Partridge Portfolio Manager: Julie Bai Field Marketing Manager: Demetrius Hall Product Marketing Manager: Yvonne Vannatta Marketing Assistant: Jon Bryant Content Managing Producer, ECS and Math: Scott Disanno Content Producer: Michelle Bayman Project Manager: Rose Kernan Assistant Project Editor, Global Editions: Vikash Tiwari Operations Specialist: Maura Zaldivar-Garcia Manager, Rights and Permissions: Ben Ferrini Senior Manufacturing Controller, Global Editions: Trudy Kimber Media Production Manager, Global Editions: Vikram Kumar Cover Designer: Lumina Datamatics Cover Photo: CT image—© zhuravliki.123rf.com/Pearson Asset Library; Gram-negative bacteria—© royaltystockphoto.com/ Shutterstock.com; Orion Nebula—© creativemarc/Shutterstock.com; Fingerprints—© Larysa Ray/Shutterstock.com; Cancer cells—© Greenshoots Communications/Alamy Stock Photo MATLAB is a registered trademark of The MathWorks, Inc., Apple Hill Drive, Natick, MA 01760-2098 Pearson Education Limited Edinburgh Gate Harlow Essex CM20 2JE England and Associated Companies throughout the world Visit us on the World Wide Web at: www.pearsonglobaleditions.com © Pearson Education Limited 2018 The rights of Rafael C Gonzalez and Richard E Woods to be identified as the authors of this work have been asserted by them in accordance with the Copyright, Designs and Patents Act 1988 Authorized adaptation from the United States edition, entitled Digital Image Processing, Fourth Edition, ISBN 978-0-13-335672-4, by Rafael C Gonzalez and Richard E Woods, published by Pearson Education © 2018 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, without either the prior written permission of the publisher or a license permitting restricted copying in the United Kingdom issued by the Copyright Licensing Agency Ltd, Saffron House, 6–10 Kirby Street, London EC1N 8TS All trademarks used herein are the property of their respective owners The use of any trademark in this text does not vest in the author or publisher any trademark ownership rights in such trademarks, nor does the use of such trademarks imply any affiliation with or endorsement of this book by such owners British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library 10 ISBN 10: 1-292-22304-9 ISBN 13: 978-1-292-22304-9 Typeset by Richard E Woods Printed and bound in Malaysia DIP4E_GLOBAL_Print_Ready.indb 7/6/2017 10:55:08 AM To Connie, Ralph, and Rob and To Janice, David, and Jonathan DIP4E_GLOBAL_Print_Ready.indb 6/16/2017 2:01:57 PM This page intentionally left blank DIP4E_GLOBAL_Print_Ready.indb 6/16/2017 2:01:57 PM Contents Preface Acknowledgments 12 The Book Website 13 The DIP4E Support Packages About the Authors 14 Introduction 13 17 What is Digital Image Processing? 18 The Origins of Digital Image Processing 19 Examples of Fields that Use Digital Image Processing Fundamental Steps in Digital Image Processing 41 Components of an Image Processing System 44 Digital Image Fundamentals 23 47 Elements of Visual Perception 48 Light and the Electromagnetic Spectrum 54 Image Sensing and Acquisition 57 Image Sampling and Quantization 63 Some Basic Relationships Between Pixels 79 Introduction to the Basic Mathematical Tools Used in Digital Image Processing 83 Intensity Transformations and Spatial Filtering 119 Background 120 Some Basic Intensity Transformation Functions 122 Histogram Processing 133 Fundamentals of Spatial Filtering 153 Smoothing (Lowpass) Spatial Filters 164 Sharpening (Highpass) Spatial Filters 175 Highpass, Bandreject, and Bandpass Filters from Lowpass Filters Combining Spatial Enhancement Methods 191 DIP4E_GLOBAL_Print_Ready.indb 188 6/16/2017 2:01:57 PM Contents Filtering in the Frequency Domain 203 Background 204 Preliminary Concepts 207 Sampling and the Fourier Transform of Sampled Functions 215 The Discrete Fourier Transform of One Variable 225 Extensions to Functions of Two Variables 230 Some Properties of the 2-D DFT and IDFT 240 The Basics of Filtering in the Frequency Domain 260 Image Smoothing Using Lowpass Frequency Domain Filters 272 Image Sharpening Using Highpass Filters 284 Selective Filtering 296 The Fast Fourier Transform 303 Image Restoration and Reconstruction 317 A Model of the Image Degradation/Restoration process 318 Noise Models 318 Restoration in the Presence of Noise Only—Spatial Filtering Periodic Noise Reduction Using Frequency Domain Filtering Linear, Position-Invariant Degradations 348 Estimating the Degradation Function 352 Inverse Filtering 356 Minimum Mean Square Error (Wiener) Filtering 358 Constrained Least Squares Filtering 363 Geometric Mean Filter 367 Image Reconstruction from Projections 368 Color Image Processing Color Fundamentals 400 Color Models 405 Pseudocolor Image Processing 420 Basics of Full-Color Image Processing Color Transformations 430 DIP4E_GLOBAL_Print_Ready.indb 327 340 399 429 6/16/2017 2:01:57 PM Contents Color Image Smoothing and Sharpening 442 Using Color in Image Segmentation 445 Noise in Color Images 452 Color Image Compression 455 Wavelet and Other Image Transforms Preliminaries 464 Matrix-based Transforms 466 Correlation 478 Basis Functions in the Time-Frequency Plane Basis Images 483 Fourier-Related Transforms 484 Walsh-Hadamard Transforms 496 Slant Transform 500 Haar Transform 502 Wavelet Transforms 504 463 479 Image Compression and Watermarking 539 Fundamentals 540 Huffman Coding 553 Golomb Coding 556 Arithmetic Coding 561 LZW Coding 564 Run-length Coding 566 Symbol-based Coding 572 Bit-plane Coding 575 Block Transform Coding 576 Predictive Coding 594 Wavelet Coding 614 Digital Image Watermarking 624 Morphological Image Processing 635 Preliminaries 636 Erosion and Dilation 638 Opening and Closing 644 The Hit-or-Miss Transform 648 DIP4E_GLOBAL_Print_Ready.indb 6/16/2017 2:01:57 PM Contents Some Basic Morphological Algorithms 652 Morphological Reconstruction 667 Summary of Morphological Operations on Binary Images Grayscale Morphology 674 10 Image Segmentation 673 699 Fundamentals 700 Point, Line, and Edge Detection 701 Thresholding 742 Segmentation by Region Growing and by Region Splitting and Merging 764 Region Segmentation Using Clustering and Superpixels 770 Region Segmentation Using Graph Cuts 777 Segmentation Using Morphological Watersheds 786 The Use of Motion in Segmentation 796 11 Feature Extraction 811 Background 812 Boundary Preprocessing 814 Boundary Feature Descriptors 831 Region Feature Descriptors 840 Principal Components as Feature Descriptors 859 Whole-Image Features 868 Scale-Invariant Feature Transform (SIFT) 881 12 Image Pattern Classification 903 Background 904 Patterns and Pattern Classes 906 Pattern Classification by Prototype Matching 910 Optimum (Bayes) Statistical Classifiers 923 Neural Networks and Deep Learning 931 Deep Convolutional Neural Networks 964 Some Additional Details of Implementation 987 Bibliography 995 Index 1009 DIP4E_GLOBAL_Print_Ready.indb 7/12/2017 10:23:39 AM www.downloadslide.net 1006 Bibliography Tizhoosh, H R [2000] “Fuzzy Image Enhancement: An Overview,” in Fuzzy Techniques in Image Processing, E Kerre and M Nachtegael (eds.), Springer-Verlag, New York Toro, J and Funt, B [2007] “A Multilinear Constraint on Dichromatic Planes for Illumination Estimation,” IEEE Trans Image Process., vol 16, no 1, pp 92–97 Tou, J T and Gonzalez, R C [1974] Pattern Recognition Principles, Addison-Wesley, Reading, MA Trussell, H J and Vrhel, M J [2008] Fundamentals of Digital Imaging, Cambridge University Press, Cambridge, UK Umbaugh, S E [2010] Digital Image Processing and Analysis, 2nd ed., CRC Press, Boca Raton, FL Vetterli, M and Kovacevic, J [1995] Wavelets and Suband Coding, Prentice Hall, Englewood Cliffs, NJ Vincent, L [1993] “Morphological Grayscale Reconstruction in Image Analysis: Applications and Efficient Algorithms,” IEEE Trans Image Process., vol no 2, pp 176–201 Walsh, J L [1923] “A Closed Set of Normal Orthogonal Functions,” Am J Math., vol 45, no 1, pp 5–24 Wang, Ruye [2012] Introduction to Orthogonal Transforms, Cambridge University Press, New York Weinberger, M J., Seroussi, G., and Sapiro, G [1996] “The LOCO-I Lossless Image Compression Algorithm: Principles and Standardization into JPEG-LS,” IEEE Trans Image Process., vol 9, no 8, pp 1309–1324 Welch, T.A [1984] “A Technique for High-Performance Data Compression,” IEEE Computer, vol 17, no 6, pp 8–19 White, J M and Rohrer, G D [1983] “Image Thresholding for Optical Character Recognition and Other Applications Requiring Character Image Extraction,” IBM J Res Devel., vol 27, no 4, pp 400–411 Whittaker, R T [1998] “A Level-Set Approach to 3D Reconstruction from Range Data,” Int’l J Comp Vision, vol 29, no 3, pp 203–231 Widrow, B and Stearns, S D [1985] Adaptive Signal Processing, Prentice Hall, Englewood Cliffs, NJ Wiener, N [1942] Extrapolation, Interpolation, and Smoothing of Stationary Time Series, MIT Press, Cambridge, MA Witten, I H., Neal, R M., and Cleary, J G [1987] “Arithmetic Coding for Data Compression,” Comm ACM, vol 30, no 6, pp 520–540 Wolberg, G [1990] Digital Image Warping, IEEE Computer Society Press, Los Alamitos, CA Woods, R E and Gonzalez, R C [1981] “Real-Time Digital Image Enhancement,” Proc IEEE, vol 69, no 5, pp 643–654 Xiang, Z., Zou, Y., Zhou, X., and Huang, X [2016] “Robust Vehicle Logo Recognition Based on Locally Collaborative Representation with Principal Components,” Sixth Int’l Conference on Information Sci and Tech., Dalian, China, pp 487-491 DIP4E_GLOBAL_Print_Ready.indb 1006 7/12/2017 10:43:36 AM www.downloadslide.net Bibliography 1007 Xintao, D., Yonglong, L., Liping, S., and Fulong, C [2014] “Color Balloon Snakes for Face Segmentation,” Int’l J for Light and Electron Optics, vol 126, no 11, pp 2538–2542 Xu, C and Prince, J, L [1988] “Snakes, Shapes, and Gradient Vector Flow,” IEEE Trans Image Process., vol.7, no 3, pp 359–369 Yu, H., Barriga, E.S., Agurto, C., Echegaray, S., Pattichis, M.S., Bauman, W., and Soliz, P [2012] “Fast Localization and Segmentation of Optic Disk in Retinal Images Using Directional Matched Filtering and Level Sets,” IEEE Trans Information Tech and Biomedicine, vol 16, no 4, pp 644 – 657 Zadeh, L A [1965] “Fuzzy Sets,” Inform and Control, vol 8, pp 338–353 Zadeh, L A [1973] “Outline of New Approach to the Analysis of Complex Systems and Decision Processes,” IEEE Trans Systems, Man, Cyb., vol SMC-3, no 1, pp 28–44 Zadeh, L A [1976] “A Fuzzy-Algorithmic Approach to the Definition of Complex or Imprecise Concepts,” Int J Man-Machine Studies, vol 8, pp 249–291 Zahn, C T and Roskies, R Z [1972] “Fourier Descriptors for Plane Closed Curves,” IEEE Trans Comput., vol C-21, no 3, pp 269–281 Zhang, Q and Skjetne, R [2015] “Image Processing for Identification of Sea-Ice Floe Size Distribution,” IEEE Trans Geoscience and Remote Sensing, vol 53, no 5, pp 2913–2924 Ziv, J and Lempel, A [1977] “A Universal Algorithm for Sequential Data Compression,” IEEE Trans Info Theory, vol IT-23, no 3, pp 337–343 Ziv, J and Lempel, A [1978] “Compression of Individual Sequences Via VariableRate Coding,” IEEE Trans Info Theory, vol IT-24, no 5, pp 530–536 DIP4E_GLOBAL_Print_Ready.indb 1007 7/12/2017 10:43:36 AM www.downloadslide.net This page intentionally left blank DIP4E_GLOBAL_Print_Ready.indb 6/16/2017 2:01:57 PM www.downloadslide.net Index A Accumulator cells, 738 Acquisition See Image: acquisition Activation functions activation value, 945 hyperbolic tangent, 945 of perceptron, 937 ReLU, 945 sigmoid, 944 Adaptive filters See Spatial filters Adjacency, 115, Affine transformations See Geometric transformations Aliasing, 234 aliased pair, 221 anti-aliasing, 222 anti-aliasing filtering, 236 examples of, 223, 234, 236, 238 in 1-D functions, 221 in images, 233 moiré patterns, 238 spatial, 233 temporal, 234 Angiogram, 27, 130 Apodizing function, 257 See also Filtering window Arctangent, 208 Arithmetic coding, 561 Artificial intelligence, 18 Artificial neurons See also Neural networks activation function, 944 activation value, 945 interconnecting, 945 model, 943 ReLU, 945 schematic, 944 sigmoid, 945 sum-of-products computation, 944 Augmented pattern See Pattern Autocorrelation, 915 B Background flattening, 689 See also Shading correction Backpropagation See Neural networks Band-limited function, 217, 225, 232 See also Function Bartlane cable system, 19 Basic rectangle, 832 Basis functions in the time-frequency plane, 479 of the cosine transform, 489 of the Fourier transform, 485 of the Haar transform, 504 of the Hartley transform, 486 of the sine transform, 493 of the slant transform, 502 of the Walsh-Hadamard transform, 499 standard, 484 Basis images, 483 See Basis functions Basis vectors, 480 See also Basis functions DIP4E_GLOBAL_Print_Ready.indb 1009 biorthonormal, 474 complex orthonormal, 473 Bayes classifier See Bayes classifier Bayes classifier See also Pattern classification as a minimum-distance classifier, 927 comparison with neural networks, 964 derivation, 923 for Gaussian pattern classes, 925 special cases, 927 Between-class variance, 748, 749 Bias See Neural networks Bidirectional frame (B-frame), 600 Biorthogonal basis, 465 Biorthonormal basis, 465 Bit-plane decomposition, 131 reconstruction, 132 Bit-plane coding, 575 Bit rate, 550 Blind spot, 49 Block matching, 600 Block transform coding, 576 bit allocation for, 583 selection of transform for, 577 subimage size and, 582 threshold implementation, 585 zonal implementation, 584 BMP compression, 566 Border, 81 See also Boundary inner, outer, 81 Boundary See also Border definition, 81 following, 814 outer, 815 resampling, 817 tracing See Boundary following Boundary descriptors, 832 basic rectangle, 832 bounding box, 832 diameter, 832 eccentricity, 832 Fourier descriptors, 835 longest chord, 832 major axis, 832 shape numbers, 834 statistical moments, 839 tortuosity, 832 Bounding box, 832 Brightness, 400 Bubble chamber, 807 C Cartesian product, 95 CAT See Computed tomography (CT) CCITT, 551 CCITT compression, 568 Ceiling function, 68, 168, 727 Cellular complex, 822–825 Center ray, 387 Chain codes first difference, 817 Freeman, 816 normalized, 817 slope, 819 Classification See Pattern classification Clustering k-means, 770 seeds, 771 supervised, 770 unsupervised, 770 Code See also Compression arithmetic, 561 CCITT makeup, 568 CCITT terminating, 568 Elias gamma, 560 Golomb, 556 Gray, 576 Huffman, 548 instantaneous, 555 length, 540 MH (Modified Huffman), 568 MMR (Modified Modified READ), 569, 574 MR (Modified READ), 569 natural binary, 542 Rice, 557 symbols, 540 unary, 557 uniquely decodable, 556 variable-length, 542 words, 540 Codec, 549 Coding See also Compression N-largest, 587 previous pixel, 596 redundancy, 540, 542 symbol-based (or token-based), 572 Cohen-Daubechies-Feauveau biorthogonal wavelets, 525, 615 Color image processing achromatic light, 400 brightness, 400 chromaticity, 403 chromaticity diagram, 404 chromatic light, 400 CMY color model, 408 CMYK color model, 408 CMYK to CMY conversion, 410 CMY to RGB conversion, 409 color circle, 434 color complements, 434 color fundamentals, 400 color “gradient”, 450 color image compression, 455 color models, 405 color pixel, 407 color slicing, 436 color transformations, 430 device independent color model, 418 edge detection, 450 full-color image processing, 429 gray level, 400 HSI color model, 411 HSI image segmentation, 446 HSI to RGB conversion, 415 6/16/2017 2:18:16 PM www.downloadslide.net 1010 Index Color Image Processing (cont.) image segmentation, 445 intensity, 400 intensity to color transformations, 423 L*a*b* color model, 419 luminance, 400 noise in color images, 452 primary colors of light, 401 primary colors of pigments, 402 pseudocolor image processing, 420 radiance, 400 RGB color cube, 408 RGB color model, 407 RGB color vector, 429, 446 RGB image segmentation, 446 RGB to CMY conversion, 409 RGB to HSI conversion, 413 secondary colors of light, 402 secondary colors of pigments, 402 sharpening, 442 smoothing, 442 tone and color corrections, 437 Complex conjugate, 207 Complex numbers, 207 angle, 208 magnitude, 207 Compression arithmetic coding, 540 bit-plane coding, 575 block diagram, 550 block transform coding, 576 BMP, 566 CCITT, 568 containers for, 551 fidelity criteria, 540 formats for, 551 fundamentals of, 540 Golomb coding, 551 Huffman coding, 551, 553 irrelevant information and, 544 JBIG2, 573 JPEG, 588 JPEG-2000, 618 Lempil-Ziv-Welch (LZW) coding, 564 measuring information for, 551 models for, 549 MPEG-4 AVC (or H.264), 603 predictive coding, 594 quantization and, 603 ratio, 540 run-length coding, 566 spatial redundancy, 603 standards for, 551 symbol-based coding, 572 temporal redundancy, 603 wavelet coding, 614 Computed tomography (CT), 21 See also Tomography Computer, history of, 21 Computerized axial tomography (CAT) See Computed tomography (CT) Conditional average risk See Bayes classifier Conditional loss See Bayes classifier Connectionist models See Neural networks Contour See Border; See Boundary Contrast See also Enhancement enhancement radiography, 27 high, 134 DIP4E_GLOBAL_Print_Ready.indb 1010 manipulation, 126 measure, 851 ratio, 69 simultaneous, 54 stretching, 121, 129 Control (tie) points, 104 Convex See also Morphological image processing deficiency, 657 digital set, 657 hull, 657 ConvNets See Neural networks (Convolutional) Convolution (Frequency domain), 213, 228 circular, 228, 253, 256 theorem, 318 tie between spatial and frequency domain filtering, 254 Convolution (Spatial) 2-D, discrete, defined, 159 and correlation, 154 computational advantage with separable kernel, 162 filter, 160 kernel, 160 kernel rotation explained, 159 mask, 160 mechanics of, 155–160 padding, 159 properties, 160 simple example of, 158 size of, 159 tie between spatial and frequency domain filtering, 254 with separable kernel, 161 Convolution theorem, 214, 215, 216, 225, 228, 253, 259, 309, 310, 312 Co-occurrence matrix See Region descriptors Correlation, 478 See also Matching coefficient, 915 image, 915 maximum, 916 minimum, 916 range of values, 916 single-point, 478 template matching See Matching Correlation (Frequency domain), 257, 258 theorem, 259 Correlation (Spatial) 2-D, discrete, defined, 158 and convolution, 154 mechanics of, 155–158 padding, 158 properties of, 160 size of, 159 Cost function, 954 Covariance matrix diagonal, 862 eigenvalues of, 861 eigenvectors of, 861 Cross-correlation, 915 Cross-modulation, 518 CT See Computed tomography (CT) Curvature, 832 estimation, 889 local, 889 obtained using Hessian matrix, 889 of edges, 889 Cutoff frequency, 274 D Data compression, 540 See also Compression Daubechies wavelets, 525, 615 dc component, 250 Decision boundary, 912 Decision function, 912 for Bayes’ classifier, 924 for minimum-distance classifier, 912 for perceptron, 935 linear, 935 Deep learning See Learning Deep neural networks See Neural networks Delta modulation (DM), 607 Denoising See Image restoration Derivative See Gradient; See also Laplacian Derivatives See also Finite differences behavior in images, 714 requirement of approximations, 702 sensitivity to noise, 714 Description regional See Regional descriptors DFT See Discrete Fourier transform; Fourier transform Difference of Gaussians, 270 Difference operators See Gradient operators Differential pulse code modulation (DPCM), 609 Digital angiography, 27 boundary, 81 computer, 21 curve, 80 filter See Filters image See Image image, definition of, 18 image representation, 65 path, 80 radiography, 25 Digital image processing See Image; See also Image fundamentals of, 47 origins of, 19 scope of, 18 steps in, 41 system components, 44 uses of, 23 Dilation See Morphological image processing Discrete cosine transform (DCT), 487 See also JPEG compression and lowpass filtering, 494 blocking artifact, 582 periodicity of, 491 Discrete Fourier transform (DFT), 475 See also Fourier transform circular convolution See Convolution circular correlation See Correlation Discrete sine transform (DST), 492 and lowpass filtering, 489 periodicity, 493 Discrete wavelet transform (DWT) See Wavelets; See also Wavelets one-dimensional, 512 two-dimensional, 520 Discriminant function See Decision function 6/16/2017 2:18:16 PM www.downloadslide.net Index Distance, 82 chessboard, 82 city-block, 82 D4, 82 D8, 82 Dm (mixed), 83 Euclidean, 82 function, 82 Mahalanobis, 926 metric, 82 Distance transform, 829 Dot product, 107 Dots per inch (dpi), 238 DPI (dots per inch), 71 Dynamic range, 69, 134 E Edge, 81 See also Edge detection closed loops, 727 derivative properties, 706 detector, definition, 702 direction, 716 double, 706, 707 global processing with Hough transform, 737 magnitude, 716 map, 721 operators See Gradient operators pixels, 702 point-linking, 735 ramp, 704, 712 roof, 704, 712 segments, 702 step, 704, 711 strength, 716 unit normal, 717 Edge detection basic steps, 714 behavior in noise, 714 Canny edge detector, 729 edge models See Edge models gradient See Gradient gradient based, 716 Laplacian of a Gaussian, 725 Marr-Hildreth detector, 724, 885 three steps performed, 714 Edge models behavior in noise, 714 ideal edges, 711 ramp edges, 712 roof edges, 712 zero crossing, 712 Eigenvalue, defined, 861 Eigenvalues See Covariance matrix Eigenvector, defined, 861 Electromagnetic (EM) spectrum, 23, 54 Elementwise operations, 83 Ellipse in standard form, 841 Encoding See Compression End point See also Morphological image processing definition, 665 Engineered features, 904, 913 Enhancement See Image enhancement Entropy, 546 Epoch See Training, epoch DIP4E_GLOBAL_Print_Ready.indb 1011 Erosion See Morphological image processing Euclidean See Distance measures distance, 82 See also Distance space, 464 vector norm, 108 Euler formula, 845 Euler number, 844 Euler’s formula, 207 Expansions linear, 467 Eye structure, 48 F Face recognition, 868 False color See Pseudocolor image processing False contouring, 74 False negative, 731 False positive, 731 Fast Fourier transform (FFT) See Discrete Fourier transform (DFT) Fast Fourier Transform (FFT) algorithm, 304 background, 204 computational advantage, 205, 269, 308 Fast wavelet transform (FWT), 513 analysis filter bank, 515 approximation and detail coefficients, 520 synthesis filter bank, 519 two-dimensional, 520 Feature extraction chain codes, 816–820 See also Chain codes corner detection, 869 covariant, 812 distance transform, 828 feature description, defined, 812 feature detection, defined, 812 feature space, 814 feature vector, 814 global, 813 invariant, 812 label, 812 local, 813 maximally stable extremal regions, 876–883 medial axis, 828 minimum-perimeter polygons, 821 Moore boundary tracing, 815 principal components, 859–869 See also Principal components region feature descriptors, 840–859 See also Region descriptors SIFT (scale invariant feature transform), 881–897 signatures, 826 skeletons, 828 whole image features, 868–899 Fidelity criteria, 547 Fiducial marks, 104, 342 Filtering (Frequency domain), 203 anti-aliasing, 236 bandpass, 297 bandreject, 297 basics of, 260 blind deconvolution, 352 computational advantage of, 205 correspondence with spatial filtering, 268, 1011 271 deblurring and denoising using constrained least squares filtering, 363 deblurring and denoising using Wiener filtering, 358 deconvolution, 352 foundation of, 215, 254 fundamentals, 261 high-boost, 291 high-frequency emphasis, 291 highpass (image sharpening), 284 highpass, offset, 263 homomorphic, 293 inverse filtering, 356 Laplacian, 289 linear, 254 lowpass (smoothing), 264, 272 notch, 299 periodic noise reduction ny notch filtering, 340 selective, 296 steps summary, 266 tie to spatial filtering, 254, 268, 271 to reduce aliasing, 222 unsharp masking, 291 Filtering (Spatial domain), 153–197 and convolution, 159 based on first derivative, 176 based on second derivative, 176 based on the gradient, 184 bias, 165, 181, 198 compared to frequency domain filters, 162 deconvolution, 352 denoising using mean filters, 328 for restoration, 327 fundamentals, 153 highboost, 182 highpass, 175 linear, 154, 160 linear, equation of, 159 lowpass, 164–174 nonlinear, 174 sharpening, 175 smoothing, 164–174 tie to frequency domain filtering, 254, 268, 271 unsharp masking, 182 Filtering windows, 257 See also Apodizing, Windowing Bartlett, 257 Hamming, 257 Hann, 257 Filtering windows (See also Apodizing) Hamming, 382 Hann, 382 Hanning, 382 Ram-Lak, 382 ramp, 382 Filters finite impulse response (FIR), 517 frequency domain See Frequency domain filtering kernels See Spatial filters spatial See Spatial filtering; See Spatial filters Filters, digital perfect reconstruction, 518 6/16/2017 2:18:16 PM www.downloadslide.net 1012 Index Filters (Frequency domain) anti-aliasing, 236 bandpass, 297 bandreject, 297 Butterworth highpass from lowpass, 284 Butterworth lowpass, 278 constrained least squares filter, 363 cutoff-frequency, 274 deconvolution, 352 difference of Gaussians, 270 filter transfer function, 261, 268 finite impulse response (FIR), 269 Gaussian highpass from lowpass, 284 Gaussian lowpass, 264, 277 geometric mean filter, 367 highpass, 262 homomorphic, 295 ideal highpass (from lowpass), 284 ideal lowpass, 220, 265, 273 inverse filter, 356 lowpass, 262 notch, 299 notch filter, 341 obtaining from spatial kernel, 271 optimum notch filter, 345 parametric Wiener filter, 367 reconstruction filters, 220 sharpening (highpass), 284 smoothing, 272 spectrum equalization, 368 Wiener filter, 358 zero-phase-shift, 266 Filters (Spatial domain) See also Kernels adaptive local noise filter, 336 adaptive median filter, 338 alpha-trimmed filter, 332 arithmetic mean filter, 328 bandpass from lowpass, 188 bandrejecct from lowpass, 188 contraharmonic mean filter, 329 geometric mean filter, 328 harmonic mean filter, 329 highpass, 175 highpass from lowpass, 188 kernel, 154 Laplacian, 178 linear, 153 max, 175 max filter, 332 median, 174 median filter, 330 midpoint filter, 332 min, 175 filter, 332 nonlinear, 153, 174 sharpening, 175 Fingerprints, 32 Finite differences, 702 backward, 703 central, 703 forward, 703 fourth order, 704 second order, 703 third order, 703 Fisher iris pattern classes, 907 Flat-field correction, 173 Floor function, 68, 168, 727 Fourier DIP4E_GLOBAL_Print_Ready.indb 1012 phase angle, 249 power spectrum, 249 spectrum, 249 Fourier descriptors See Boundary descriptors Fourier series, 475 defined, 208 history, 204 Fourier-slice theorem See Image reconstruction from projections Fourier transform, 252 1-D continuous, 210 1-D discrete (DFT), 225, 226, 227 1-D inverse discrete (IDFT), 227 2-D, continuous, 231 2-D discrete (DFT), 240 and aliasing, 222 and convolution, 214 center, 243, 259 centering, 242 computing inverse from forward, 304 conditions for existence, 210 convolution See Convolution dc component, 250 derivation of DFT, 225 discrete See Discrete Fourier transform (DFT) forward, 210, 227, 228, 272 history, 204 inverse, 227 inverse, 1-D continuous, 210 inverse from forward algorihm, 304 magnitude, 249 of 2-D box, 231 of impulse, 208 of impulse and impulse train, 212 of pulse, 211 of sampled functions, 216 pair, 214, 227, 231 periodicity, 241 phase angle, 249 power spectrum, 249 properties of, 240, 246, 258 separability, 303 spectrum, 249 symmetry, 243 Four-quadrant arctangent, 207, 249 Fractals, 39 Frequency domain, defined, 211 rectangle, 243 units of, 211 Frequency domain filtering See also Spatial filtering; See Filtering (Frequency domain) Frequency leakage, 257 Function antisymmetric, 244 band-limited See Band-limited function Bessel, 276 circularly symmetric, 167 complex, 243 conjugate symmetric, 245 Dirac delta, 208 even, 243 frequency of a periodic, 223 generalized, 208 isotropic, 167 modulation, 346 odd, 243 period, 223 periodic, 223 real, 243, 245 reconstruction (recovery), 224 sinc, 212 symmetric, 244 weighting, 346 Fundus image, 833 G Gamma correction, 125 noise See Noise Gaussian See also Probability density function (PDF) circularly symmetric shape, 167 convolution of Gaussians, 169 coveriance matrix See also Covariance matrix difference of (DoG), 727, 885 first derivative of, 729 isotropic shape, 167 kernel, 883 kernel max size needed, 168 See also Kernels Laplacian of (LoG), 725 noise, 86, 319 noise, white, 729 product of Gaussians, 169 Gaussian noise See Noise Generalized eigenvalue problem, 783 Geometric transformations, 100 Global thresholding See Thresholding Golden image, 116 Golomb codes and coding, 556 Gradient combined with thresholding, 722 definition of, 184, 716 direction, 716 kernels, 718–721 See also Kernels magnitude, 185, 716 operators, 717 vector, 716 Gradient descent, 940 Granular noise, 608 Granulometry, 685 Graph cuts (for segmentation) See Graph cuts directed, 778 nodes, 778 undirected, 778 vertices, 778 weighted, 779 Graph cuts, 777 See also Image segmentation computing, 783 max-flow, min-cut theorem, 781 minimum, 780 normalized cut, 782 segmentation algorithm, 785 Gray level, 18, 57, 400 See also Intensity Gray-level transformation See Intensity transformations Gray scale, 63 See also Intensity Ground truth, 929 6/16/2017 2:18:16 PM www.downloadslide.net Index H Haar functions, 502 Haar transform, 502, 522 Haar wavelets, 615 Hadamard product, 83, 490 Halftone dots, 238 Harris matrix, 871 Harris-Stephens corner detector, 869 Hartley transform, 485 and lowpass filtering, 494 Heisenberg-Gabor inequaltiy, 481 Heisenberg uncertainty principle, 481 Hertz, 211 Hessian matrix, 888 Histogram bins, 133 equalization See Histogram processing linearization See Histogram, equalization normalized, 133 specification See Histogram processing, matching statistics, 150 unnormalized, 133 Histogram processing, 140 equalization, 134 local, 149, 152 matching, 140 using statistics, 150 Homogeneity, 84 Hotelling transform See Principal components Hough transform, 737 Huffman coding, 553 Human eye See Visual perception Hyperplanes, 912 Hyperspheres, 927 I IDFT See Fourier transform, 227 IEC, 551 Illumination, 57 See also Intensity bias, 763 effect on thresholding, 745 sinusoidal, 764 spot, 763 Image, 27 acquisition, 57, 58, 60, 62 analysis, 18 arithmetic, 85 background, 81, 97, 743 bit planes, 131 center, 68 classification See Pattern classification columns, 67 comparing, 87 complement, 99 compression See Compression contrast, 69 contrast ratio, 69 coordinates, 18 correspondence between (x, y) and (row, col), 68 denoising See Image restoration difference, 87 DIP4E_GLOBAL_Print_Ready.indb 1013 dynamic range, 69 element, 66 See Pixel EM spectrum, 23 enhancement See Enhancement foreground, 81, 97 formation model, 61 gamma-ray, 24 golden, 116 grayscale, 57 illumination, 62 See Illumination indexing See Indexing intensity, 18 See Intensity interpolation, 77, 237 See Interpolation masking, 90 mathematical tools, 83–118 microwave, 33 morphology See Morphological image processing MRI, 34 multispectral, 29 negative, 99, 123 neighborhood, 120 origin, definition, 67 padding, 157, 158 patch, 870 PET, 24 pixel, 18 See Pixel quantization, 63 radar, 33 radio, 34 recognition See Pattern classification reference, 103 reflectance, 62 registration, 103, 106 resampling, 237 resizing, 237 resolution See Resolution restoration See Image restoration rotation See Geometric transformations rows, 67 sampling, 63 saturation, 69 scaling See Geometric transformations sensing, 57 sharpening, 175 shearing See Geometric transformations shrinking, 237 smoothing, 164 sound, 36 spatial coordinates, 18 synthetic, 23 transformation kernel, 109 translation See Geometric transformations ultrasound, 38 ultraviolet, 27 understanding, 18 watermarking, 624 X-ray, 24 zooming, 237 Image enhancement character repair, 282 combined spatial methods, 191 contrast enhancement, 126 contrast stretching, 129 defined, 122 gamma correction, 125 gradient based, 184 1013 highboost filtering, 182, 291 high-frequency emphasis, 291 homomorphic, 293 intensity-level slicing, 130 intensity transformations, 122 interference reduction, 283, 300, 302 Laplacian based, 178, 289 local, 151, 152 mask mode radiography, 89 median filtering, 174 noise reduction by averaging, 86 shading correction, 90, 173 sharpening, 284 sharpening and thresholding, 289 smoothing, 164, 272, 282 thresholding, 173 unsharp masking, 182, 291 ImageNet Challenge, 933 Image quantization, 63 Image reconstruction from projections absorption profile, 368 backprojections, 369, 371, 377, 387, 388 backprojections, fan-beam filtered, 386 backprojections, parallel-beam filtered, 380, 384 center ray, 386 CT principles, 370 CT scanner generations, 372 filtered backprojection, 383 Fourier-slice theorem, 379, 380 introductory example, 368 parallel-beam filtered backprojections, 380 parallel-ray beam projection, 374 Radon transform, 374, 375 Radon transform example, 376 Ram-Lak filter, 382 raysum, 368, 374 Shepp-Logan phantom, 377 sinogram, 377 slice, 380 smearing, 368 windowing, 382, 385 Image registration, 103 Image restoration adaptive filtering, 333 adaptive mean filter, 338 alpha-trimmed filtering, 332 arithmetic mean filter, 328 blind deconvolution, 352 constrained least square filtering, 363 constrained least squares filtering, 363 contraharmonic mean filter, 329 deblurring, 361, 362, 364 deblurring by inverse filtering, 361 deblurring using a least squares filter, 364 deblurring using a Wiener filter, 364 deconvolution, 352 degradation function, 352 degradation function estimation, 352 denoising using adaptive median filtering, 339 denoising using linear spatial filters, 327 denoising using notch filtering, 342 denoising using order-statistic filtering, 333 denoising using spatial mean filters, 329 6/16/2017 2:18:16 PM www.downloadslide.net 1014 Index Image restoration (cont.) Erlang (Gamma) noise model, 321 exponential noise model, 321 Gaussian noise model, 319 geometric mean filtering, 367 harmonic mean filter, 329 image blurring caused by motion, 356 interactive, 367 interference removal using notch filters, 347 linear, position invariant degradations, 348 max filtering, 332 median filtering, 330 midpoint filtering, 332 filtering, 332 modeling motion blurr, 354 model of, 318 noise reduction in the frequency domain, 340 notch filtering, 341 optimum noise filtering, 345 order-statistic filters, 330 periodic noise model, 324 Rayleigh noise nodel, 320 salt-and-pepper noise model, 322 uniform noise model, 321 Wiener filtering, 358 Image sampling, 63, 215 Image segmentation clustering (K-means), 770 cluster seeds, 771 definition, 700 edge-based, 710–742 edge-point linking, 735 finite differences, 702 graph cuts See Graph cuts line detection, 707 morphological watersheds, 786 motion-based, 796 point detection, 706 region growing, 764 region splitting and merging, 768 seeds, 764 superpixels, 772 textural segmentation, 687 texture based, 770 thresholding See Thresholding unsupervised clustering, 770 Image transforms, 109 See Transforms Impulse (Frequency domain), 208 1-D continuous, 208 1-D discrete, 209 2-D continuous, 230 Fourier transform of, 212 sifting property, 208, 210, 230 train, 212, 216 unit, 208, 259 Impulse response, 269, 350 Impulse (Spatial domain) strength, 157 unit discrete, 157 Independent frames (I-frames), 599 Indexing, 70 coordinate, 70 linear, 70 Information theory, 545 Inner product, 107 Intensity, 18, 63, 67, 68 discrete, 68 DIP4E_GLOBAL_Print_Ready.indb 1014 discrimination, 53 levels, 70, 73 range, 68 resolution, 71, 72 scale, 63 scaling, 91 slicing, 130 variance See also Moments Intensity mappings See Intensity transformations Intensity transformations definition, 121 for histogram equalization, 134, 138 monotonic, 135 piecewise-linear, 128 single valued, 136 thresholding, 121 Interior angle, 822 Interpolation, 77, 101, 225 bicubic, 78 bilinear, 77 control points, 104 nearest neighbor, 77 splines, 79 tie points, 104 wavelets, 79 Inverse proportionality, 211, 213, 229, 241, 276 Inverse transforms See Transforms Iris data (Fisher), 906 classification using min-dist-classifier, 912 classification using perceptrons, 939 ISO, 551 Isolated point definition See also Image segmentation Isopreference curves, 76 ITU-T, 551 J JBIG2 compression, 573 JPEG-2000 compression, 618 irreversible component transform, 618 lifting-based wavelet transform for, 619 tile components, 619 JPEG compression, 588 K Karhunen-Loève transform See Principal components Kernels bandpass from lowpass, 188 bandreject from lowpass, 188 box (lowpass), 165 defined, 154 edge detection, 717 for implementing gradient, 184 Gaussian, 160 Gaussian lowpass, 166, 727 highpass from lowpass, 188 how to construct, 164 in terms of lowpass kernels, 189 isotropic, 708, 725 Kirsch compass, 720 Laplacian, 179 line detection, 707 normalizing, 168 point detection, 706 Prewitt, 718 Roberts, 185, 718 separable, 109, 161 separable, computational advantage, 162 separable, construction of, 162 Sobel, 187, 719 symmetric, 109 transformation, 109 K-means clustering See Image segmentation L Laminogram See Image reconstruction from projections Laplacian defined, 179 kernel See Kernels line detection, 708 sharpening, 180 Laplacian of a Gaussian See Edge detection Learning See also Pattern classification; Neural networks deep, 905, 931 defined, 905 history of, 931 increment, 935 machines, 932 over fitting, 905, 929 rate, 935 supervised, 905 unsupervised, 905 Leftmost uppermost point, 815 Leibniz’s rule, 136 Lempel-Ziv-Welch (LZW) coding, 564 Lens (of eye), 49 Light, 54 See also Electromagnetic (EM) spectrum achromatic, 57, 400 chromatic, 57, 400 color, 57 intensity, 57 luminance, 57 monochromatic, 57 photon, 56 radiance, 57 wavelength, 56 white, 57 Line definition See also Image segmentation normal representation, 374, 738 slope-intercept form, 374, 738 Linear additivity property, 349 convolution See Convolution correlation See Correlation homogeneity property, 349 index See Indexing operation, 348 operator, 84 transform pair, 109 transforms, 109 Linearly separable, 935 Linear system, 350 See also Linear operator characterization by impulse, 350 6/16/2017 2:18:16 PM www.downloadslide.net Index convolution integral, 351 superposition (Fredholm) integral, 350 Line pairs, 71 Lloyd-Max quantizer, 613 LMSE algorithm See Perceptrons Logical operations, 96 AND, 97 FALSE, 97 functionally complete, 97 NOT (complement), 97 OR, 97 TRUE, 97 XOR, 97 Lookup table, 122, 144 Luminance, 57, 400 LZW coding See Lempel-Ziv-Welch (LZW) coding M Mach bands, 53 Macroblock, 599 Magnetic Ink Character Recognition (MICR), 914 Mahalanobis distance, 448, 758, 926 See also Distance measures Major axis, 832 Mapper, 550 inverse, 604 Mapping See Intensity: mapping Markov sources, 547 Mask mode radiography, 89 Masks See Spatial filters threshold, 586 zonal, 584 Matching See also Pattern classification correlation, 915 degree of similarity, 920 prototypes, 910 shape numbers, 919 SIFT features, 917 similarity matrix, 920 similarity tree, 920 strings, 920 template, 915 Matrix covariance See Covariance matrix product, 84 rank, 162 Matrix determinant in terms of eigenvalues, 872 Matrix trace in terms of eigenvalues, 872 Maximally stable extremal regions See Feature extraction Maximally stable extremal regions (MSER’s) See Feature extraction Mean absolute distortion (MAD), 600 Mean filters See Spatial filters Mean of intensity See Moments Mean squared error, 939 See also Neural networks Mean-squared signal-to-noise ratio, 548 Medial axis, 828 Mexican hat operator, 725 Microdensitometers, 58 Microscopy electron, 39 DIP4E_GLOBAL_Print_Ready.indb 1015 fluorescence, 27 light, 28 Minimum-distance classifier, 910 decision function, 912 Minimum-perimeter polygons See Feature extraction Minkowsky addition, 693 Minkowsky subtraction, 693 Minor axis, 832 MNIST, 979 Modulation, 518 Moiré patterns, 238 Moment invariants, 858 Moments See also Population global mean, 151 global variance, 151 local mean, 151 local variance, 151 Monotonic function, 135 strict, 135 Moore boundary tracing algorithm See Feature extraction Morphological algorithms See Morphological image processing Morphological image processing algorithms (binary), 652 background flattening, 689 background pixels, 636, 638 border clearing, 672 See Morphological reconstruction bottom-hat transformation, 684 boundary extraction, 653 broken character repair, 642 closing, 644 closing by reconstruction, 670 closing by reconstruction, grayscale, 689 closing, grayscale, 680 closing, properties, 648 complement, 644, 646 computationally efficient, 667 conditional operations, 654, 667 connected component extraction, 655 convex deficiency, 657 convex digital set, 657 convex hull, 657 detecting foreign objects in packaged food, 656 dilation, 641 dilation, grayscale, 678 don’t-care elements, 651 don’t-care elements, 636, 639, 665, 673 duality, 644, 647 end point detection, 665 erosion, 639 erosion, grayscale, 674 filtering, 641, 648, 682 foreground pixels, 636, 638 geodesic dilation, 667 geodesic dilation, grayscale, 688 geodesic erosion, 667 geodesic erosion, grayscale, 689 gradient, 682 granulometry, 685 grayscale images, 673 hit-or-miss transform, 648 hole, defined, 651, 653 hole filling, 653, 671 marker image, 689 1015 mask image, 689 opening, 644 opening by reconstruction, 670 opening by reconstruction, grayscale, 689 opening, grayscale, 680 opening, properties, 648 pruning, 664 reconstruction, 667 See Morphological reconstruction reconstruction by dilation, 668 reconstruction by dilation, grayscale, 689 reconstruction by erosion, 669 reconstruction by erosion, grayscale, 689 reconstruction, grayscale, 688 reflection, set, 636 rolling ball analogy, 645 segmentation, textural, 687 segmentation using watersheds, 786 SE See Structuring element, 636 shading correction, 685 skeletons, 662 See also Skeletons smoothing, 682 spurs, 664 structuring element, defined, 636 structuring element, isotropic, 667 structuring element, nonflat, 678 structuring elements, examples of, 637 summary of binary operations, 673 surface area, 686 thickening, 660 thinning, 660 thinning, topology-preserving, 828 top-hat by reconstruction, 689 top-hat transformation, 683 translation, set, 637 Motion and its use in segmentation, 796 See also Image segmentation Motion compensation, predictive coding and, 599 MSER’s See Maximally stable extremal regions Multiresolution analysis, 506 N National Institute of Standards and Technology, 979 Nearest-neighbor classifier See Minimumdistance classifier Neighbor 4-neighbor, 79 8-neighbor, 79 diagonal, 79 nearest See also Interpolation north, south, east, west, 815 Neighborhood, 79 closed, 79 definition, 120 open, 79 operations, 99, 120 Neural networks activation functions See Activation functions backpropagation, 932, 953, 974 convolutional See Neural networks (Convolutional) deep, 906, 931, 945, 946 6/16/2017 2:18:17 PM www.downloadslide.net 1016 Index Neural networks (cont.) deep learning, 932, 933 design considerations, 987 error (classification), 959 error (mean squared) (MSE), 959 feedforward, 945 fully-connected See Neural networks (Fully-connected) multilayer, 943 over-fitting, 981, 989 pretrained, 989 shallow, 945 using GPUs, 989 Neural networks (Convolutional) activation functions See Activation functions activation value, 945 artificial neuron See Artificial neurons background, 931 backpropagation derivation, 974 backpropagation equations, 974 basic architecture, 964 bias, 964 CNN for large character dataset, 979 CNN for large image database, 982 convolutional layer, 966 convolution equations, 975 dropout, 989 feature map, 966 forward pass equations, 973 graphical illustration of functioning, 968 implementation details, 987 kernel, 966 mean squared error, 978, 980 mini-batches,, 989 multiple input images, 973 neural computations, 971 overfitting, 989 padding, 988 pooled feature maps, 967 pooling, 966 pooling methods, 967 pooling neighborhoods, 967 receptive fields, 966 recognition error, 978, 981, 985 rot180 (rotation), 975 rotated kernel, 975 stochastic gradient, 989 stride, 966 subsampling, 966 summary table, 977 sum of products, 936, 937 training by backpropagation, 974 updsampling, 977 vanishing gradients, 988 vectorizing, 968 visual cortex model, 966 weight (parameter) sharing, 966 Neural networks (Fully-connected) activation functions See Activation functions activation value, 948 artificial neuron See Artificial neurons background, 931 backpropagation, derivation, 953 backpropagation, matric formulation, 956 bias, 935, 944, 947 DIP4E_GLOBAL_Print_Ready.indb 1016 chain rule, 953 comparison with Bayes’ classifier, 964 comparison with the Bayes’ classifier, 964 correction increment, 935 deep, 945 dropout, 989 error function, 954 feedforward, 945 forward pass, equations of, 948 forward pass, matrix formulation, 950 hidden layers, 945 implementation details, 987 input vectors, 948 learning See Learning learning increment, 935 learning rate, 935 mean squared error, 961 mini-batches,, 989 model of, 946 multilayer, 943 multispectral data classification, 961 net input, 948 overfitting, 989 padding, 988 pretrained network, 989 recognition error, 978 shallow, 946 stochastic gradient, 989 sum of products, 937 training, 953 training by backpropagation, 953 training epoch See Epoch upsampling, 988 vanishing gradients, 988 weights, 946 XOR problem solution, 959 Neurocomputers See Neural networks Neuromorphic systems See Neural networks NIST, 979 Noise autocorrelation of, 359 bipolar impulse, 322 data-drop-out, 322 density, 322 Erlang (Gamma), 321 Exponential, 321 Gaussian, 319 granular, 608 impulse bipolar, 322 impulse unipolar, 322 parameter estimation, 325 periodic, 324, 340 Rayleigh, 320 salt-and-pepper, 322 spike, 322 uniform, 321 unipolar, 322 white, 319, 360, 729 white Gaussian, 729 Noiseless coding theorem, 546 Nonlinear operator, 84 Nonmaxima suppression, 730 Normalized central moments, 858 Notch filters See Frequency domain filtering Nyquist rate, 219 See also Sampling; See also Sampling O Objective function, 954 Object recognition See Patterns: recognition and Opening See Morphological image processing Operations arithmetic, 85 elementwise, 83 linear, 84 logical, 91, 96 matrix, 84, 106 neighborhood, 99 nonlinear, 84 set, 91 spatial, 98 vector, 106 Operator linear, operator See Linear operator position invariance, 349 space invariance, 349 Optical illusions, 54 Order partial, 96 strict, 96 Ordered pairs, 95 See also Cartesian product Order-statistic filters See Spatial filters origins of, 19 Orthogonal basis, 465 Orthonormal basis, 465 Otsu’s method See Threshold; See Thresholding Outer product, 107 Over fitting See Learning Over-segmentation, 793 P Padding, 155, 158, 256 and periodicity, 264 difference between spatial and frequency, 265 illustration, 171, 172 mirror, 171 replicate, 171 size, 159, 256, 266 symmetric, 171 zero, 165, 166, 171, 256, 265 Parameter space, 738 Path, 80, 83 Pattern augmented, 936 classes, 906 definition of, 904 formation, 906 labeling, 905 recognition See Pattern classification space, 906 strings, 910 test, 905 training, 905 trees, 910 unlabeled, 905 validation, 905 vectors, 906, 909 6/16/2017 2:18:17 PM www.downloadslide.net Index Pattern classification See also Learning approaches, 904 Bayes’ classifier See Classifier correlation, 915 decision functions, 912 See also Decision functions deep learning See Learning discriminant functions., 912 feature extraction See Feature extraction hyperplane, 912 minimum-distance classifier, 910 pattern See Pattern pattern class, definition of, 904 pattern classes, 906 pattern recognition, stages, 904 pattern strings See Pattern pattern tree See Pattern pattern vectors, 906 prototype matching See Matching SIFT See Matching string matching See Matching supervised learning See Learning test set, 905 training set, 905 unsupervised learning, 905 validation set, 905 Pattern recognition See also Pattern classification stages of, 904 PDF See Probability density function (PDF) Pel See Pixel Perceptrons bias, 935 convergence theorem, 936 correction increment, 935 decision boundary, 938 decision surfaces, 934 hyperplanes, 935 learning rate, 935 least-mean-squared-error (LMSE), 941 linearly separable See Linearly separable parameters, 935 schematic, 937 sum of products computation, 937 training algorithm, 935 training epoch, 937 weights, 935 weight vector, 935 XOR problem, 942 Photons, 23, 56 Pixel adjacency of, 80 connected, 80 definition, 18, 66 foreground See Image, background; See Image, foreground interpolation See Interpolation neighborhood operations See also Spatial filtering neighbors of, 79 object, 743 replication, 237 transformation See Intensity: transformations Point processing, 121 Point spread function, 350 DIP4E_GLOBAL_Print_Ready.indb 1017 Pointwise and max operators, 688 Population See also Moments, Sample Positron emission tomography, 24 See also Tomography Predicate region growing, 766 thresholding, 762 Prediction errors, 597 Prediction residuals, 597 motion compensated, 599 Predictive coding, 594 delta modulation (DM), 607 differential pulse code modulation (DPCM), 609 lossless, 594 lossy, 605 motion compensation and, 599 optimal predictors for, 609 optimal quantization in, 611 prediction error, 594 Predictive frame (P-frame), 599 Prewitt gradient operators See Spatial filters Principal components covariance matrix, 861 eigenvalues, 861 eigenvectors, 861 Hotelling transform, 861 Karhunen-Loève transform, 861 mean vector, 861 rotation normalization, 865 size normalization, 865 transformation, 861 translation normalization, 865 vector formation, 863 Probability a priori, 923 cumulative distribution function (CDF) See Cumulative distribution function (CDF) density function (PDF) See also Probability density function mass function (PMF) See also Probability mass function prior, 923 Probability density function (PDF) Erlang (Gamma), 321 Exponential, 321 Gaussian, 319 Rayleigh, 320 Salt-and-Pepper, 322 Uniform, 321 Probability mass function (PMF), 558 Probability models, 562 Projection-slice theorem See Image reconstruction from projections Pruning See Morphological image processing Q Quad tree, 768 Quantization, 545 See also Sampling Quantizer, 550 inverse, 604 Lloyd-Max, 613 optimum uniform, 614 1017 R Radar band, 33 Radiance, 57, 400 Radon transform formulation, 374 history, 371 Ramp edges See Edges Random variable See also Probability Ranger 7, 21 Rayleigh noise See Noise Redundancy, 540 coding, 540, 541 relative data, 540 spatial and temporal, 540, 543 Region adjacent, 700 border, 81 boundary, 81 contour, 81 definition, 80 descriptors See Description disjoint, 700 Region descriptors circularity, 840 compactness, 840 connected components, 844 contrast, 851 co-occurrence matrix, 849 correlation, 850 eccentricity, 841 effective diameter, 841 entropy, 851 Euler number, 844 holes, 844 homogeneity, 851 major and minor axes, 840 moments, 846 polygonal networks, 845 principal axes, 841 roundness, 840 texture (spectral), 855 texture (statistical), 846 topological, 843 uniformity, 848, 851 Relation antireflexive, 96 binary, 96 ordering, 96 transitive, 96 Remote sensing, 29 LANDSAT, 29 thematic bands, 29 Resampling See Image: resampling Reseau marks, 21, 104 Resolution, 71 dots per inch (dpi), 71 intensity, 71 line pairs, 71 spatial, 71 Root-mean-squared (rms) error, 360, 548 Rubber-sheet distortions, 843 Run-length encoding (RLE), 566 Run-length pairs, 543, 566 6/16/2017 2:18:17 PM www.downloadslide.net 1018 Index S Salt-and-pepper noise See Noise Sample See also Population Sampling, 63, 215, 233 See also Quantization aliasing See Aliasing critically-sampled signal, 217 limitations of human eye, 236 Nyquist rate, 219 of printed material, 238 over-sampled signal, 217 rate, 215, 217, 223, 232 relationship to frequency intervals, 228 theorem, 217, 231, 236 undersampled signal, 221 units of sampling rate, 215 Saturation, 69 Scale-invariant feature transform See SIFT Scale space See SIFT Scaling geometric See Geometric transformations intensity, 91 Scaling functions, 505 coefficients of, 507 Haar, 505 refinement equation for, 507 scale and translation of, 505 separable, 520 Self-adaptive networks See Neural networks Sensors acquisition and, 57 arrays, 61 single, 58 strips, 60 sequency, 497 Sets, 91, 93 complement, 92 difference, 92 intersection, 92 mutually exclusive, 92 on grayscale images, 94 partially-ordered, 96 sample space, 92 set universe, 92 strict-ordered, 96 union, 92 Sets and set operations, 644 reflection, 636 translation, 637 Shading correction, 90, 685 See also Background flattening using lowpass filtering, 173 Shannon’s first theorem, 546 Sharpening frequency See Filtering (Frequency domain) spatial See Filtering (Spatial domain) Shepp-Logan phantom See Image reconstruction from projections Shrinking See Image: resampling SIFT algorithm, 894 curvature estimation using Hessian matrix, 889 examples of image matching, 894–897 Hessian matrix, 888 DIP4E_GLOBAL_Print_Ready.indb 1018 keypoint descriptors, 892 keypoint locations accuracy, 887 keypoint orientation, 890 keypoints, 882 local extrema, 885 octaves, 883 scale parameter, 883 scale space, 883 subpixel accuracy, 888 Sifting property See Impulse Signal-to-noise ratio, 360 Signatures, 826 Sinc function, 212 See also Function Sinogram See Image reconstruction from projections Skeletons, 828 See also Morphological image processing defined, 829 via the distance transform, 828 Slant transform, 500 Sliding inner product, 478 Slope density function, 827 Slope overload, 608 SMPTE, 551 Sobel gradient operators See Spatial filtering Softmax function, 947 See also Neural networks Space constant, 724 Spaghetti effect, 727 Spatial coordinates, 66 definition, 18 transformation of, 101 Spatial domain convolution See Convolution correlation See Correlation definition, 119 filtering See Spatial filtering Spatial filtering See Filtering (Spatial domain) masks See Spatial filters Spatial filters See also Spatial filtering Spatial operations, 98 Spatial variables, 66 Spectrum See Discrete Fourier transform (DFT); See Fourier transform Statistical moments See Moments Step edges See Edges Strict ordering, 96 Structured light, 31 Structuring element See Morphological image processing Subband coding, 517 Subpixel accuracy, 78, 727, 888 Subspace analysis tree, 527 Successive-doubling See Fast Fourier transform Sum of absolute distortions (SAD), 600 Sum of products, 154, 159, 213 Superpixels as graph nodes, 779 defined, 772 SLIC algorithm, 774 Superposition, 350 Supervoxels, 777 Symlets, 525, 615 Symmetry antihermitian, 245 conjugate, 245 conjugate antisymmetry, 245 even, 249, 364 hermitian, 245 odd, 249 T Taylor series, 702 Texture See also Region descriptors spectral, 855 statistical, 846 Thickening See Morphological image processing Thinning See Morphological image processing Threshold See also Thresholding coding, 584 Thresholding adaptive, 743 basics, 743 document, 763 dynamic, 743 global, 746 hysteresis, 732, 759 illumination, role of, 745 local, 743 multiple, 757 multivariate, 448 noise, role of, 744 Otsu’s method, 747 predicate, 762 reflectance, role of, 745 regional, 743 using edges, 753 using smoothing, 752 variable, 761 Tie (control) points, 104 Time-Frequency Plane, 479 Token, 572 Tomography, 368 See also Image reconstruction from projections magnetic resonance imaging (MRI), 374 single photon emission, (SPECT), 374 X-ray computed tomography, 370 X-ray CT scanner generations, 372 Training defined, 905 epoch, 937, 958 neural networks See Neural networks over-fitting See Neural networks patterns See Pattern perceptrons See Perceptrons test set, 905 training set, 905 validation set, 905 vanishing gradients, 988 Training by backpropagation, 953, 974 Transformation affine, 101 geometric (rubber-sheet) See Geometric transformations kernel, 467 matrix, 468 6/16/2017 2:18:17 PM www.downloadslide.net Index matrix dual, 474 rotate, 101 scale, 101 sheer, 101 translate, 101 Transformation functions bit-plane slicing, 131 Transforms cosine, 487, 577 Fourier, 577 See Fourier transform Haar, 502 Hartley, 485 Hough See Hough transform matrix-based, 466 morphological See Morphological image processing orthogonal, 470 sine, 492 slant, 500 Walsh-Hadamard, 496, 577 wavelet, 504 See also Wavelets U Uniform See Noise Unitary matrix, 473 space, 464 transform, 473 universality approximation theorem universality approximation theorem, 987 Uppermost leftmost point, 815 V Vanishing gradient, 988 Vanishing moments, 524 Variable thresholding See Thresholding Variance of intensity See Moments Vector operations angle between vectors, 465 dot (scalar) product, 464 inner product, 464 integral inner product, 464 DIP4E_GLOBAL_Print_Ready.indb 1019 norm, 464 orthogonal complement, 508 Vertex concave, 822 convex, 822 degenerate, 822 Video compression standards, 604 Vision See also Visual perception high-level, 18 inspection, 31 low-level, 18 machine, 18 mid-level, 18 Visual perception, 48 blind spot, 49 brightness adaptation, 51 brightness discrimination, 51 human eye structure, 48 image formation in eye, 50 Mach bands, 53 optical illusions, 54 photopic vision, 49 scotopic vision, 49 simultaneous contrast, 54 subjective brightness, 51 Voxels, 429 W Walsh functions, 496 Walsh-Hadamard transform (WHT), 496 Hadamard (natural) ordering, 496 sequency ordering, 498 Watermarking digital images, 624 block diagram for, 628 reasons for, 624 Watermarks, 624 attacks on, 627 fragile invisible, 626 insertion and extraction, 627 invisible, 625 private (or restricted key), 627 public (or unrestricted key), 627 robust invisible, 627 visible, 625 1019 Wavelength, 56 of EM spectrum, 56 required to get image, 57 Wavelet coding, 614 decomposition level selection, 616 quantizer design for, 617 selection of wavelets for, 615 Wavelet functions, 507 coefficients of, 508, 525 Haar, 509 Wavelet packets, 526 filter bank for, 529 subspace analysis tree, 527 Wavelets, 479, 482 and edge detection, 524 and multiresolution analysis, 506 packets, 526 scaling functions, 505 See also Scaling functions series expansions, 510 wavelet functions, 507 See also Wavelet functions Weber ratio, 52 White noise See Noise Windowing, 257 See also Apodizing, Filtering windows Wraparound error, 256, 257, 262, 266, 271, 300 X XOR classification problem definition, 942 solution using neural nets, 959 solution using perceptrons, 943 Z Zero crossing, 178 Zero crossings, 712, 724, 726 Zero-memory source, 546 Zonal coding implementation, 584 Zone plate, 190 Zooming See Image: zooming 6/16/2017 2:18:17 PM www.downloadslide.net GLOBAL EDITION Digital Image Processing For these Global Editions, the editorial team at Pearson has collaborated with educators across the world to address a wide range of subjects and requirements, equipping students with the best possible learning tools This Global Edition preserves the cutting-edge approach and pedagogy of the original, but also features alterations, customization, and adaptation from the North American version GLOBAL EDITION Digital Image Processing FOURTH EDITION Rafael C Gonzalez • Richard E Woods FOURTH EDITION Gonzalez Woods GLOBAL EDITION This is a special edition of an established title widely used by colleges and universities throughout the world Pearson published this exclusive edition for the benefit of students outside the United States and Canada If you purchased this book within the United States or Canada, you should be aware that it has been imported without the approval of the Publisher or Author The Global Edition is not supported in the United States and Canada Pearson Global Edition Gonzalez_04_1292223049_Final.indd 11/08/17 5:27 PM ... Origins of Digital Image Processing 19 Examples of Fields that Use Digital Image Processing Fundamental Steps in Digital Image Processing 41 Components of an Image Processing System 44 Digital Image. .. Companion Website for Rafael C Gonzalez and Richard E Woods’ Digital Image Processing, Fourth Edition, Global Edition 1.  Go to www.ImageProcessingPlace.com 2.  Find the title of your textbook 3. Click... the concept of a digital image Have a broad overview of the historical underpinnings of the field of digital image processing Understand the definition and scope of digital image processing Know