Machine Vision: Davies: 9780122060922: Book: : eCampus.com

This book provides a detailed background to machine vision, a subject that has evolved to embrace a diverse range of topics. With an emphasis on the theory underpinning practicalities, the book covers the area of image processing, image analysis and machine/computer vision, including automated visual inspection. The second edition incorporates many recent advances in the theory and practice of machine vision, including 3-D interpretation, invariants, camera calibration, artificial neural networks, x-ray inspection and foreign object detection, mathematical morphology, robust statistics, and an updated and very extensive list of references.

Preface to First Edition

xxi

Preface to Second Edition

xxiv

About the Author

xxvi

Glossary of Acronyms and Abbreviations

xxvii

Acknowledgements

xxix

Vision, the Challenge

(18)

Introduction---man and his senses

(2)

The nature of vision

(8)

The process of recognition

(1)

Tackling the recognition problem

(3)

Object location

(3)

Scene analysis

(1)

Vision as inverse graphics

(1)

Automated visual inspection

(2)

What this book is about

(1)

The following chapters

(1)

Bibliographical notes

(4)

Part 1 Low-Level Processing

Images and Imaging Operations

(22)

Introduction

(3)

Grey scale versus colour

(1)

Image processing operations

(11)

Some basic operations on grey-scale images

(4)

Basic operations on binary images

(4)

Noise suppression by image accumulation

(2)

Convolutions and point spread functions

(3)

Sequential versus parallel operations

(2)

Concluding remarks

(1)

Bibliographical and historical notes

(1)

Problems

(2)

Basic Image Filtering Operations

(38)

Introduction

(2)

Noise suppression by Gaussian smoothing

(3)

Median filtering

(2)

Mode filtering

(7)

Bias generated by noise suppression filters

(11)

Theory of edge shifts caused by median filters in binary images

(5)

Edge shifts caused by median filters in grey-scale images

(3)

Edge shifts arising with hybrid median filters

(2)

Problems with statistics

(1)

Reducing computational load

(5)

A bit-based method for fast median filtering

(2)

VLSI implementation of the median filter

(1)

The role of filters in industrial applications of vision

(1)

Sharp-unsharp masking

(1)

Concluding remarks

(2)

Bibliographical and historical notes

(1)

Problems

(3)

Thresholding Techniques

(24)

Introduction

(1)

Region-growing methods

(1)

Thresholding

(12)

Finding a suitable threshold

(2)

Tackling the problem of bias in threshold selection

(2)

Methods based on finding a valley in the intensity distribution

(1)

Methods which concentrate on the peaked intensity distribution at high gradient

(2)

A convenient mathematical model

(6)

Summary

(1)

Adaptive thresholding

(5)

The Chow and Kaneko approach

(1)

Local thresholding methods

(3)

Concluding remarks

(2)

Bibliographical and historical notes

100

(1)

Problems

101

(2)

Locating Objects via Their Edges

103

(28)

Introduction

103

(1)

Basic theory of edge detection

104

(2)

The template matching approach

106

(1)

Theory of 3 x 3 template operators

107

(6)

Summary---design constraints and conclusions

113

(1)

The design of differential gradient operators

114

(2)

The concept of a circular operator

116

(1)

Detailed implementation of circular operators

117

(2)

Structured bands of pixels in neighbourhoods of various sizes

119

(4)

The systematic design of differential edge operators

123

(1)

Problems with the above approach---some alternative schemes

124

(4)

Concluding remarks

128

(1)

Bibliographical and historical notes

129

(1)

Problems

130

(1)

Binary Shape Analysis

131

(40)

Introduction

131

(1)

Connectedness in binary images

132

(1)

Object labelling and counting

133

(4)

Metric properties in digital images

137

(3)

Size filtering

140

(2)

The convex hull and its computation

142

(4)

Distance functions and their uses

146

(5)

Skeletons and thinning

151

(12)

Crossing number

153

(3)

Parallel and sequential implementations of thinning

156

(3)

Guided thinning

159

(2)

A comment on the nature of the skeleton

161

(1)

Skeleton node analysis

161

(2)

Application of skeletons for shape recognition

163

(1)

Some simple measures for shape recognition

163

(1)

Shape description by moments

164

(1)

Boundary tracking procedures

165

(2)

Concluding remarks

167

(1)

Bibliographical and historical notes

168

(1)

Problems

169

(2)

Boundary Pattern Analysis

171

(24)

Introduction

171

(2)

Boundary tracking procedures

173

(1)

Template matching---a reminder

174

(1)

Centroidal profiles

174

(2)

Problems with the centroidal profile approach

176

(5)

Some solutions

178

(3)

The (s,) plot

181

(2)

Tackling the problems of occlusion

183

(3)

Chain code

186

(1)

The (r, s) plot

187

(1)

Accuracy of boundary length measures

187

(2)

Concluding remarks

189

(1)

Bibliographical and historical notes

190

(1)

Problems

191

(4)

Part 2 Intermediate-Level Processing

Line Detection

195

(16)

Introduction

195

(1)

Application of the Hough transform to line detection

196

(4)

The foot-of-normal method

200

(7)

Error analysis

202

(2)

Quality of the resulting data

204

(1)

Application of the foot-of-normal method

205

(2)

Longitudinal line localization

207

(1)

Final line fitting

208

(1)

Concluding remarks

208

(1)

Bibliographical and historical notes

209

(1)

Problem

210

(1)

Circle Detection

211

(34)

Introduction

211

(1)

Hough-based schemes for circular object detection

212

(5)

The problem of unknown circle radius

217

(7)

Experimental results

222

(2)

The problem of accurate centre location

224

(8)

Obtaining a method for reducing computational load

226

(2)

Improvements on the basic scheme

228

(2)

Discussion

230

(1)

Practical details

230

(2)

Overcoming the speed Problem

232

(10)

More detailed estimates of speed

233

(2)

Robustness

235

(2)

Experimental results

237

(3)

Summary

240

(2)

Concluding remarks

242

(1)

Bibliographical and historical notes

242

(1)

Problem

243

(2)

The Hough Transform and Its Nature

245

(26)

Introduction

245

(1)

The generalized Hough transform

246

(2)

Setting up the generalized Hough transform---some relevant questions

248

(1)

Spatial matched filtering in images

248

(1)

From spatial matched filters to generalized Hough transforms

249

(2)

Gradient weighting versus uniform weighting

251

(4)

Calculation of sensitivity and computational load

252

(3)

Summary

255

(1)

Applying the generalized Hough transform to line detection

256

(1)

An instructive example

257

(1)

Tradeoffs to reduce computational load

258

(2)

The effects of occlusions for objects with straight edges

260

(3)

Fast implementations of the Hough transform

263

(3)

The approach of Gerig and Klein

266

(1)

Concluding remarks

267

(1)

Bibliographical and historical notes

268

(3)

Ellipse Detection

271

(20)

Introduction

271

(1)

The diameter bisection method

271

(3)

The chord-tangent method

274

(1)

Finding the remaining ellipse parameters

275

(2)

Reducing computational load for the generalized Hough transform method

277

(9)

Practical details

282

(4)

Comparing the various methods

286

(2)

Concluding remarks

288

(1)

Bibliographical and historical notes

289

(1)

Problems

290

(1)

Hole Detection

291

(18)

Introduction

291

(1)

The template matching approach

291

(2)

The lateral histogram technique

293

(1)

The removal of ambiguities in the lateral histogram technique

294

(4)

Computational implications of the need to check for ambiguities

294

(2)

Further detail of the subimage method

296

(2)

Application of the lateral histogram technique for object location

298

(5)

Limitations of the approach

302

(1)

A strategy based on applying the histograms in turn

303

(2)

Appraisal of the hole detection problem

305

(2)

Concluding remarks

307

(1)

Bibliographical and historical notes

308

(1)

Problems

308

(1)

Polygon and Corner Detection

309

(38)

Introduction

309

(1)

The generalized Hough transform

310

(1)

Straight edge detection

310

(1)

Application to the detection of regular polygons

311

(1)

The case of an arbitrary triangle

312

(1)

The case of an arbitrary rectangle

313

(2)

Lower bounds on the numbers of parameter planes

315

(4)

An extension of the triangle result

319

(1)

Discussion

320

(4)

Determining orientation

324

(1)

Why corner detection?

324

(1)

Template matching

325

(1)

Second-order derivative schemes

326

(3)

A median-based corner detector

329

(5)

Analysing the operation of the median detector

329

(3)

Practical results

332

(2)

The Hough transform approach to corner detection

334

(4)

The lateral histogram approach to corner detection

338

(2)

Corner orientation

340

(1)

Concluding remarks

340

(2)

Bibliographical and historical notes

342

(1)

Problems

343

(4)

Part 3 Application-Level Processing

Abstract Pattern Matching Techniques

347

(26)

Introduction

347

(1)

A graph-theoretic approach to object location

348

(8)

A practical example---locating cream biscuits

353

(3)

Possibilities for saving computation

356

(3)

Using the generalized Hough transform for feature collation

359

(3)

Computational load

360

(2)

Generalizing the maximal clique and other approaches

362

(1)

Relational descriptors

363

(4)

367

(1)

Concluding remarks

368

(1)

Bibliographical and historical notes

369

(1)

Problems

370

(3)

The Three-Dimensional World

373

(44)

Introduction

373

(1)

Three-dimensional vision---the variety of methods

374

(2)

Projection schemes for three-dimensional vision

376

(8)

Binocular images

378

(2)

The correspondence problem

380

(4)

Shape from shading

384

(5)

Photometric stereo

389

(2)

The assumption of surface smoothness

391

(2)

Shape from texture

393

(1)

Use of structured lighting

394

(2)

Three-dimensional object recognition schemes

396

(2)

The method of Ballard and Sabbah

398

(3)

The method of Silberberg et al.

401

(1)

Horaud's junction orientation technique

402

(5)

The 3DPO system of Bolles and Horaud

407

(2)

The IVISM system

409

(2)

Lowe's approach

411

(2)

Concluding remarks

413

(1)

Bibliographical and historical notes

414

(2)

Problems

416

(1)

Tackling the Perspective N-Point Problem

417

(14)

Introduction

417

(1)

The phenomenon of perspective inversion

417

(2)

Ambiguity of pose under weak perspective projection

419

(3)

Obtaining unique solutions to the pose problem

422

(6)

Solution of the three-point problem

426

(1)

Using symmetric trapezia for estimating pose

427

(1)

Concluding remarks

428

(1)

Bibliographical and historical notes

429

(2)

Motion

431

(24)

Introduction

431

(1)

Optical flow

431

(4)

Interpretation of optical flow fields

435

(2)

Using focus of expansion to avoid collision

437

(2)

Time-to-adjacency analysis

439

(2)

Basic difficulties with the optical flow model

441

(1)

Stereo from motion

441

(3)

Applications to the monitoring of traffic flow

444

(7)

The system of Bascle et al.

444

(3)

The system of Koller et al.

447

(4)

Concluding remarks

451

(1)

Bibliographical and historical notes

452

(3)

Invariants and Their Applications

455

(16)

Introduction

455

(2)

Cross ratios: the ``ratio of ratios'' concept

457

(4)

Invariants for noncollinear points

461

(4)

Invariants for points on conics

465

(2)

Concluding remarks

467

(1)

Bibliographical and historical notes

468

(3)

Automated Visual Inspection

471

(32)

Introduction

471

(1)

The process of inspection

472

(1)

Review of the types of object to be inspected

473

(3)

Food products

473

(1)

Precision components

474

(1)

Differing requirements for size measurement

475

(1)

Three-dimensional objects

475

(1)

Other products and materials for inspection

476

(1)

Summary---the main categories of inspection

476

(2)

Shape deviations relative to a standard template

478

(1)

Inspection of circular products

479

(7)

Computation of the radial histogram: statistical problems

482

(3)

Application of radial histograms

485

(1)

Inspection of printed circuits

486

(2)

Steel strip and wood inspection

488

(1)

Inspection of products with high levels of variability

488

(5)

X-ray inspection

493

(4)

Bringing inspection to the factory

497

(2)

Concluding remarks

499

(1)

Bibliographical and historical notes

500

(3)

Statistical Pattern Recognition

503

(26)

Introduction

503

(1)

The nearest neighbour algorithm

504

(3)

Bayes' decision theory

507

(3)

Relation of the nearest neighbour and Bayes' approaches

510

(3)

Mathematical statement of the problem

510

(3)

The importance of the nearest neighbour classifier

513

(1)

The optimum number of features

513

(1)

Cost functions and error-reject tradeoff

514

(2)

Cluster analysis

516

(6)

Supervised and unsupervised learning

516

(1)

Clustering procedures

517

(5)

Principal components analysis

522

(3)

The relevance of probability in image analysis

525

(1)

Concluding remarks

526

(1)

Bibliographical and historical notes

527

(1)

Problems

528

(1)

Biologically Inspired Recognition Schemes

529

(32)

Introduction

529

(1)

Artificial neural networks

530

(6)

The back-propagation algorithm

536

(3)

MLP architectures

539

(1)

Overfitting to the training data

540

(3)

Optimizing the network architecture

543

(1)

Hebbian learning

544

(5)

Case-study: noise suppression using ANNs

549

(5)

Genetic algorithms

554

(3)

Concluding remarks

557

(1)

Bibliographical and historical notes

558

(3)

Texture

561

(22)

Introduction

561

(3)

Some basic approaches to texture analysis

564

(2)

Grey-level co-occurrence matrices

566

(3)

Laws' texture energy approach

569

(3)

Ade's eigenfilter approach

572

(2)

Appraisal of the Laws and Ade approaches

574

(2)

Fractal-based measures of texture

576

(1)

Shape from texture

577

(1)

Markov random field models of texture

578

(1)

Structural approaches to texture analysis

579

(1)

Concluding remarks

579

(1)

Bibliographical and historical notes

580

(3)

Image Acquisition

583

(20)

Introduction

583

(1)

Illumination schemes

584

(8)

Eliminating shadows

586

(3)

Arranging a region of uniform illumination

589

(1)

Use of linescan cameras

590

(2)

Cameras and digitization

592

(4)

Digitization

595

(1)

The sampling theorem

596

(4)

Concluding remarks

600

(1)

Bibliographical and historical notes

600

(3)

The Need for Speed: Real-Time Electronic Hardware Systems

603

(20)

Introduction

603

(1)

Parallel processing

604

(2)

SIMD systems

606

(2)

The gain in speed attainable with N processors

608

(1)

Flynn's classification

609

(2)

Optimal implementation of an image analysis algorithm

611

(4)

Hardware specification and design

613

(1)

Basic ideas on optimal hardware implementation

613

(2)

Board-level processing systems

615

(1)

VLSI

616

(1)

Concluding remarks

617

(1)

Bibliographical and historical notes

618

(5)

Part 4 Perspectives on Vision

Machine Vision, Art or Science?

623

(72)

Introduction

623

(1)

Parameters of importance in machine vision

624

(2)

Tradeoffs

626

(3)

Some important tradeoffs

627

(1)

Tradeoffs for two-stage template matching

628

(1)

Future directions

629

(2)

Hardware, algorithms and processes

631

(1)

A retrospective view

631

(2)

Just a glimpse of vision?

633

(1)

Bibliographical and historical notes

633

(2)

Appendices

Appendix A: Programming Notation

635

(12)

A.1 Introduction

635

(1)

A.2 The Pascal language

636

(1)

A.2.1 Control structures

636

(2)

A.2.2 Procedures and functions

638

(1)

A.2.3 Other details of Pascal syntax

639

(3)

A.2.4 The need for special syntax

642

(1)

A.3 Special syntax embedded in Pascal

642

(1)

A.3.1 Image handling notation

642

(1)

A.3.2 Other succinct notation

643

(3)

A.4 On the validity of the ``repeat until finished'' construct

646

(1)

Appendix B: Mathematical Morphology

647

(16)

B.1 Introduction

647

(1)

B.2 Dilation and erosion in binary images

648

(1)

B.2.1 Dilation and erosion

648

(1)

B.2.2 Cancellation effects

648

(1)

B.2.3 Modified dilation and erosion operators

649

(1)

B.3 Mathematical morphology

649

(1)

B.3.1 Generalized morphological dilation

649

(2)

B.3.2 Generalized morphological erosion

651

(1)

B.3.3 Duality between dilation and erosion

652

(3)

B.3.4 Closing and opening

655

(1)

B.3.5 Hit-and-miss transform

656

(1)

B.3.6 Template matching

657

(1)

B.4 Connectedness-based analysis of images

658

(1)

B.4.1 Skeletons and thinning

658

(2)

B.5 Concluding remarks

660

(1)

B.6 Bibliographical and historical notes

660

(3)

Appendix C: Image Transformations and Camera Calibration

663

(14)

C.1 Introduction

663

(1)

C.2 Image transformations

663

(5)

C.3 Camera calibration

668

(3)

C.4 Intrinsic and extrinsic parameters

671

(4)

C.5 Concluding remarks

675

(1)

C.6 Bibliographical and historical notes

675

(2)

Appendix D: Robust statistics

677

(18)

D.1 Introduction

677

(3)

D.2 Preliminary definitions and analysis

680

(3)

D.3 The M-estimator (influence function) approach

683

(5)

D.4 The least median of squares approach to regression

688

(3)

D.5 Overview of the robustness problem

691

(2)

D.6 Concluding remarks

693

(1)

D.7 Bibliographical and historical notes

694

(1)

References

695

(30)

Subject Index

725

(16)

Author Index

741


	Buy Textbooks Sell Textbooks College Apparel Shop by School Virtual Bookstores	Order Status Shipping Rates Return Policy Marketplace Info F.A.S.T.	Contact Us Privacy Policy Legal Notices Site Security Employment	Help Desk eCampus Blog Affiliate Program Bulk Orders College Marketing

Need Help? eService@ecampus.com Copyright© 1999-2008
.