机器视觉理论、算法与实践(英文版·第3版)(图灵原版计算机科学系列)(Machine Vision Thcory，Algorithms，Practicalities Third Edition)
点此进入淘宝搜索页搜索分类: 图书,计算机与互联网,计算机科学理论,计算机结构、设计与制造,
品牌: E.R.Davies
基本信息·出版社：人民邮电出版社
·页码：934 页
·出版日期：2009年
·ISBN：7115195498/9787115195494
·条形码：9787115195494
·包装版本：3版
·装帧：平装
·开本：16
·正文语种：中文
·丛书名：图灵原版计算机科学系列
·外文书名：Machine Vision Thcory，Algorithms，Practicalities Third Edition
产品信息有问题吗？请帮我们更新产品信息。
内容简介《机器视觉理论、算法与实践(英文版·第3版)》是机器视觉课程的理想教材，作者清晰、系统地阐述了机器视觉的基本概念，介绍理论的基本元素的同时强调算法和实用设计的约束。书中阐述各个主题时，既阐述了基本算法，又介绍了数学工具。此外，《机器视觉理论、算法与实践(英文版·第3版)》还使用案例演示具体技术的应用，并阐明设计现实机器视觉系统的关键约束。
《机器视觉理论、算法与实践(英文版·第3版)》适合作为高等院校计算机及电子工程相关专业研究生的教材，更是从事机器视觉、计算机视觉和机器人领域研究的人员不可多得的技术参考书。
作者简介E.R.Davies，著名机器视觉专家。英国物理学会会士、IEE会士、英国机器视觉协会的执行委员。毕业于牛津大学，现任伦敦大学皇家霍洛威学院机器视觉教授。在机器视觉、图像分析、自动视觉检测、噪声抑制技术等方面有丰富的教学和科研经验。
媒体推荐“本书将图像处理的理论与应用实践完美地结合起来，是机器视觉领域研究人员的必读之作。”
——John Billingsley，南昆士兰大学
“前两版已经奠定了本书在机器视觉领域中独一无二的地位，它是对重要的图像处理和计算机视觉算法进行详细分析的知识宝库！这一版在此基础之上增加了最新进展，是一部全面而且与时俱进的权威著作。”
——Farzin Deravi，肯特大学
编辑推荐40年来，机器视觉在各行各业得到了广泛的应用，包括自动检测、机器人组装、行车导引、流量监控、签名验证、生物测量、遥感图像分析等。但是另一方面，面对大量新的研究成果，要充分理解相关的理论和应用，进行算法和系统的设计，却越来越困难。
《机器视觉理论、算法与实践(英文版·第3版)》能够满足广大读者学习和掌握机器视觉知识的需求。全书图文并茂，清晰、系统地阐述了基本概念，提供了丰富的应用案例和代码，强调了算法和实用设计的各种约束条件。新版做了全面的更新，反映了最新进展，内容更加全面。《机器视觉理论、算法与实践(英文版·第3版)》是机器视觉课程的理想教材，已经成为国内外很多名校的指定教学参考书。同时，《机器视觉理论、算法与实践(英文版·第3版)》也是工程技术人员不可或缺的权威参考书。
目录
CHAPTER 1Vision, the Challenge
1.1Introduction-The Senses1
1.2The Nature of Vision2
1.2.1The Process of Recognition2
1.2.2Tackling the Recognition Problem4
1.2.3Object Location7
1.2.4Scene Analysis9
1.2.5Vision as Inverse Graphics10
1.3From Automated Visual Inspection to Surveillance11
1.4What This Book Is About12
1.5The Following Chapters14
1.6Bibliographical Notes15
PART 1LOW-LEVELVISION17
CHAPTER 2Images and Imaging Operations
2.1Introduction19
2.1.1Gray-scale versus Color21*
2.2Image Processing Operations24
2.2.1Some Basic Operations on Gray-scale Images25
2.2.2Basic Operations on Binary Images32
2.2.3Noise Suppression by Image Accumulation37
2.3Convolutions and Point Spread Functions39
2.4Sequential versus Parallel Operations41
2.5Concluding Remarks43
2.6Bibliographical and Historical Notes44
2.7Problems44
CHAPTER 3Basic Image Filtering Operations
3.1Introduction47
3.2Noise Suppression by Gaussian Smoothing49
3.3Median Filters51
3.4Mode Filters54
3.5Rank Order Filters61
3.6Reducing Computational Load61
3.6.1A Bit-based Method for Fast Median Filtering64
3.7Sharp-Unsharp Masking65
3.8Shifts Introduced by Median Filters66
3.8.1Continuum Model of Median Shifts68
3.8.2Generalization to Gray-scale Images72
3.8.3Shifts Arising with Hybrid Median Filters75
3.8.4Problems with Statistics76
3.9Discrete Model of Median Shifts78
3.9.1Generalization to Gray-scale Images82
3.10Shifts Introduced by Mode Filters84
3.11Shifts Introduced by Mean and Gaussian Filters86
3.12Shifts Introduced by Rank Order Filters86
3.12.1Shifts in Rectangular Neighborhoods87
3.12.2Case of High Curvature91
3.12.3Test of the Model in a Discrete Case91
3.13The Role of Filters in Industrial Applications of Vision94
3.14Color in Image Filtering94
3.15Concluding Remarks96
3.16Bibliographical and Historical Notes96
3.17Problems98
CHAPTER 4Thresholding Techniques
4.1Introduction103
4.2Region-growing Methods104
4.3Thresholding105
4.3.1Finding a Suitable Threshold105
4.3.2Tackling the Problem of Bias in Threshold Selection107
4.3.3A Convenient Mathematical Model111
4.3.4Summary114
4.4Adaptive Thresholding114
4.4.1The Chow and Kaneko Approach118
4.4.2Local Thresholding Methods119
4.5More Thoroughgoing Approaches to Threshold Selection122
4.5.1Variance-based Thresholding122
4.5.2Entropy-based Thresholding123
4.5.3Maximum Likelihood Thresholding125
4.6Concluding Remarks126
4.7Bibliographical and Historical Notes127
4.8Problems129
CHAPTER 5Edge Detection
5.1Introduction131
5.2Basic Theory of Edge Detection132
5.3The Template Matching Approach133
5.4Theory of 3×3 Template Operators135
5.5Summary-Design Constraints and Conclusions140
5.6The Design of Differential Gradient Operators141
5.7The Concept of a Circular Operator143
5.8Detailed Implementation of Circular Operators144
5.9Structured Bands of Pixels in Neighborhoods of Various Sizes146
5.10The Systematic Design of Differential Edge Operators150
5.11Problems with the above Approach-Some Alternative Schemes151
5.12Concluding Remarks155
5.13Bibliographical and Historical Notes156
5.14Problems157
CHAPTER 6Binary Shape Analysis
6.1Introduction159
6.2Connectedness in Binary Images160
6.3Object Labeling and Counting161
6.3.1Solving the Labeling Problem in a More Complex Case164
6.4Metric Properties in Digital Images168
6.5Size Filtering169
6.6The Convex Hull and Its Computation171
6.7Distance Functions and Their Uses177
6.8Skeletons and Thinning181
6.8.1Crossing Number183
6.8.2Parallel and Sequential Implementations of Thinning186
6.8.3Guided Thinning189
6.8.4A Comment on the Nature of the Skeleton189
6.8.5Skeleton Node Analysis191
6.8.6Application of Skeletons for Shape Recognition192
6.9Some Simple Measures for Shape Recognition193
6.10Shape Description by Moments194
6.11Boundary Tracking Procedures195
6.12More Detail on the Sigma and Chi Functions196
6.13Concluding Remarks197
6.14Bibliographical and Historical Notes199
6.15Problems200
CHAPTER 7Boundary Pattern Analysis
7.1Introduction207
7.1.1Hysteresis Thresholding209
7.2Boundary Tracking Procedures212
7.3Template Matching-A Reminder212
7.4Centroidal Profiles213
7.5Problems with the Centroidal Profile Approach214
7.5.1Some Solutions216
7.6The (s,ψ) Plot218
7.7Tackling the Problems of Occlusion220
7.8Chain Code223
7.9The (r, s) Plot224
7.10Accuracy of Boundary Length Measures225
7.11Concluding Remarks227
7.12Bibliographical and Historical Notes228
7.13Problems229
CHAPTER 8Mathematical Morphology
8.1Introduction233
8.2Dilation and Erosion in Binary Images234
8.2.1Dilation and Erosion234
8.2.2Cancellation Effects234
8.2.3Modified Dilation and Erosion Operators235
8.3Mathematical Morphology235
8.3.1Generalized Morphological Dilation235
8.3.2Generalized Morphological Erosion237
8.3.3Duality between Dilation and Erosion238
8.3.4Properties of Dilation and Erosion Operators239
8.3.5Closing and Opening242
8.3.6Summary of Basic Morphological Operations245
8.3.7Hit-and-Miss Transform248
8.3.8Template Matching249
8.4Connectivity-based Analysis of Images249
8.4.1Skeletons and Thinning250
8.5Gray-scale Processing251
8.5.1Morphological Edge Enhancement252
8.5.2Further Remarks on the Generalization to Gray-scale Processing252
8.6Effect of Noise on Morphological Grouping Operations255
8.6.1Detailed Analysis257
8.6.2Discussion259
8.7Concluding Remarks259
8.8Bibliographical and Historical Notes260
8.9Problem261
PART 2INTERMEDIATE-LEVELVISION263
CHAPTER 9Line Detection
9.1Introduction265
9.2Application of the Hough Transform to Line Detection265
9.3The Foot-of-Normal Method269
9.3.1Error Analysis272
9.3.2Quality of the Resulting Data274
9.3.3Application of the Foot-of-Normal Method276
9.4Longitudinal Line Localization276
9.5Final Line Fitting277
9.6Concluding Remarks277
9.7Bibliographical and Historical Notes278
9.8Problems280
CHAPTER 10Circle Detection
10.1Introduction283
10.2Hough-based Schemes for Circular Object Detection284
10.3The Problem of Unknown Circle Radius288
10.3.1Experimental Results290
10.4The Problem of Accurate Center Location295
10.4.1Obtaining a Method for Reducing Computational Load296
10.4.2Improvements on the Basic Scheme299
10.4.3Discussion300
10.4.4Practical Details300
10.5Overcoming the Speed Problem302
10.5.1More Detailed Estimates of Speed303
10.5.2Robustness305
10.5.3Experimental Results306
10.5.4Summary307
10.6Concluding Remarks310
10.7Bibliographical and Historical Notes311
10.8Problems312
CHAPTER 11The Hough Transform and Its Nature
11.1Introduction315
11.2The Generalized Hough Transform315
11.3Setting Up the Generalized Hough Transform-Some Relevant Questions317
11.4Spatial Matched Filtering in Images318
11.5From Spatial Matched Filters to Generalized Hough Transforms319
11.6Gradient Weighting versus Uniform Weighting320
11.6.1Calculation of Sensitivity and Computational Load323
11.7Summary324
11.8Applying the Generalized Hough Transform to Line Detection325
11.9The Effects of Occlusions for Objects with Straight Edges327
11.10Fast Implementations of the Hough Transform329
11.11The Approach of Gerig and Klein332
11.12Concluding Remarks333
11.13Bibliographical and Historical Notes334
11.14Problem337
CHAPTER 12Ellipse Detection
12.1Introduction339
12.2The Diameter Bisection Method339
12.3The Chord-Tangent Method341
12.4Finding the Remaining Ellipse Parameters343
12.5Reducing Computational Load for the Generalized Hough Transform Method345
12.5.1Practical Details349
12.6Comparing the Various Methods353
12.7Concluding Remarks355
12.8Bibliographical and Historical Notes357
12.9Problems358
CHAPTER 13Hole Detection
13.1Introduction361
13.2The Template Matching Approach361
13.3The Lateral Histogram Technique363
13.4The Removal of Ambiguities in the Lateral Histogram Technique363
13.4.1Computational Implications of the Need to Check for Ambiguities364
13.4.2Further Detail of the Subimage Method366
13.5Application of the Lateral Histogram Technique for Object Location368
13.5.1Limitations of the Approach370
13.6Appraisal of the Hole Detection Problem372
13.7Concluding Remarks374
13.8Bibliographical and Historical Notes375
13.9Problems376
CHAPTER 14Polygon and Corner Detection
14.1Introduction379
14.2The Generalized Hough Transform380
14.2.1Straight Edge Detection380
14.3Application to Polygon Detection381
14.3.1The Case of an Arbitrary Triangle382
14.3.2The Case of an Arbitrary Rectangle383
14.3.3Lower Bounds on the Numbers of Parameter Planes385
14.4Determining Polygon Orientation387
14.5Why Corner Detection?389
14.6Template Matching390
14.7Second-order Derivative Schemes391
14.8A Median-Filter-Based Corner Detector393
14.8.1Analyzing the Operation of the Median Detector394
14.8.2Practical Results396
14.9The Hough Transform Approach to Corner Detection399
14.10The Plessey Corner Detector402
14.11Corner Orientation404
14.12Concluding Remarks406
14.13Bibliographical and Historical Notes407
14.14Problems410
CHAPTER 15Abstract Pattern Matching Techniques
15.1Introduction413
15.2A Graph-theoretic Approach to Object Location414
15.2.1A Practical Example-Locating Cream Biscuits419
15.3Possibilities for Saving Computation422
15.4Using the Generalized Hough Transform for Feature Collation424
15.4.1Computational Load426
15.5Generalizing the Maximal Clique and Other Approaches427
15.6Relational Descriptors428
15.7Search432
15.8Concluding Remarks433
15.9Bibliographical and Historical Notes434
15.10Problems437
PART 33-DVISION AND MOTION443
CHAPTER 16The Three-dimensional World
16.1Introduction445
16.2Three-Dimensional Vision-The Variety of Methods446
16.3Projection Schemes for Three-dimensional Vision448
16.3.1Binocular Images450
16.3.2The Correspondence Problem452
16.4Shape from Shading454
16.5Photometric Stereo459
16.6The Assumption of Surface Smoothness462
16.7Shape from Texture464
16.8Use of Structured Lighting464
16.9Three-Dimensional Object Recognition Schemes466
16.10The Method of Ballard and Sabbah468
16.11The Method of Silberberg et al.470
16.12Horaud’s Junction Orientation Technique472
16.13An Important Paradigm-Location of Industrial Parts476
16.14Concluding Remarks478
16.15Bibliographical and Historical Notes480
16.16Problems482
CHAPTER 17Tackling the Perspective n-Point Problem
17.1Introduction487
17.2The Phenomenon of Perspective Inversion487
17.3Ambiguity of Pose under Weak Perspective Projection489
17.4Obtaining Unique Solutions to the Pose Problem493
17.4.1Solution of the3-Point Problem497
17.4.2Using Symmetrical Trapezia for Estimating Pose498
17.5Concluding Remarks498
17.6Bibliographical and Historical Notes501
17.7Problems502
CHAPTER 18Motion
18.1Introduction505
18.2Optical Flow505
18.3Interpretation of Optical Flow Fields509
18.4Using Focus of Expansion to Avoid Collision511
18.5Time-to-Adjacency Analysis513
18.6Basic Difficulties with the Optical Flow Model515
18.7Stereo from Motion516
18.8Applications to the Monitoring of Traffic Flow518
18.8.1The System of Bascle et al.518
18.8.2The System of Koller et al.520
18.9People Tracking524
18.9.1Some Basic Techniques526
18.9.2Within-vehicle Pedestrian Tracking528
18.10Human Gait Analysis530
18.11Model-based Tracking of Animals-A Case Study533
18.12Snakes536
18.13The Kalman Filter538
18.14Concluding Remarks540
18.15Bibliographical and Historical Notes542
18.16Problem543
CHAPTER 19Invariants and Their Applications
19.1Introduction545
19.2Cross Ratios: The “Ratio of Ratios” Concept547
19.3Invariants for Noncollinear Points552
19.3.1Further Remarks about the5-Point Configuration554
19.4Invariants for Points on Conics556
19.5Differential and Semidifferential Invariants560
19.6Symmetrical Cross Ratio Functions562
19.7Concluding Remarks564
19.8Bibliographical and Historical Notes566
19.9Problems567
CHAPTER 20Egomotion and Related Tasks
20.1Introduction571
20.2Autonomous Mobile Robots572
20.3Active Vision573
20.4Vanishing Point Detection574
20.5Navigation for Autonomous Mobile Robots576
20.6Constructing the Plan View of Ground Plane579
20.7Further Factors Involved in Mobile Robot Navigation581
20.8More on Vanishing Points583
20.9Centers of Circles and Ellipses585
20.10Vehicle Guidance in Agriculture-A Case Study588
20.10.13-D Aspects of the Task590
20.10.2Real-time Implementation591
20.11Concluding Remarks592
20.12Bibliographical and Historical Notes592
20.13Problems593
CHAPTER 21Image Transformations and Camera Calibration
21.1Introduction595
21.2Image Transformations596
21.3Camera Calibration601
21.4Intrinsic and Extrinsic Parameters604
21.5Correcting for Radial Distortions607
21.6Multiple-view Vision609
21.7Generalized Epipolar Geometry610
21.8The Essential Matrix611
21.9The Fundamental Matrix613
21.10Properties of the Essential and Fundamental Matrices614
21.11Estimating the Fundamental Matrix615
21.12Image Rectification616
21.133-D Reconstruction617
21.14An Update on the8-Point Algorithm619
21.15Concluding Remarks621
21.16Bibliographical and Historical Notes622
21.17Problems623
PART 4TOWARD REAL-TIME PATTERN RECOGNITIONS YSTEMS625
CHAPTER 22Automated Visual Inspection
22.1Introduction627
22.2The Process of Inspection628
22.3Review of the Types of Objects to Be Inspected629
22.3.1Food Products629
22.3.2Precision Components630
22.3.3Differing Requirements for Size Measurement630
22.3.4Three-dimensional Objects631
22.3.5Other Products and Materials for Inspection632
22.4Summary-The Main Categories of Inspection632
22.5Shape Deviations Relative to a Standard Template634
22.6Inspection of Circular Products635
22.6.1Computation of the Radial Histogram: Statistical Problems636
22.6.2Application of Radial Histograms641
22.7Inspection of Printed Circuits642
22.8Steel Strip and Wood Inspection643
22.9Inspection of Products with High Levels of Variability644
22.10X-ray Inspection648
22.11The Importance of Color in Inspection651
22.12Bringing Inspection to the Factory653
22.13Concluding Remarks654
22.14Bibliographical and Historical Notes656
CHAPTER 23Inspection of Cereal Grains
23.1Introduction659
23.2Case Study 1: Location of Dark Contaminants in Cereals660
23.2.1Application of Morphological and Nonlinear Filters to Locate Rodent Droppings663
23.2.2Appraisal of the Various Schemas664
23.2.3Problems with Closing665
23.3Case Study 2: Location of Insects665
23.3.1The Vectorial Strategy for Linear Feature Detection666
23.3.2Designing Linear Feature Detection Masks for Larger Windows669
23.3.3Application to Cereal Inspection670
23.3.4Experimental Results671
23.4Case Study 3: High-speed Grain Location673
23.4.1Extending an Earlier Sampling Approach673
23.4.2Application to Grain Inspection675
23.4.3Summary679
23.5Optimizing the Output for Sets of Directional Template Masks680
23.5.1Application of the Formulas682
23.5.2Discussion683
23.6Concluding Remarks683
23.7Bibliographical and Historical Notes684
CHAPTER 24Statistical Pattern Recognition
24.1Introduction687
24.2The Nearest Neighbor Algorithm688
24.3Bayes’ Decision Theory691
24.4Relation of the Nearest Neighbor and Bayes’ Approaches693
24.4.1Mathematical Statement of the Problem693
24.4.2The Importance of the Nearest Neighbor Classifier696
24.5The Optimum Number of Features696
24.6Cost Functions and Error-Reject Tradeoff697
24.7The Receiver-Operator Characteristic699
24.8Multiple Classifiers702
24.9Cluster Analysis705
24.9.1Supervised and Unsupervised Learning705
24.9.2Clustering Procedures706
24.10Principal Components Analysis710
24.11The Relevance of Probability in Image Analysis713
24.12The Route to Face Recognition715
24.12.1The Face as Part of a3-D Object716
24.13Another Look at Statistical Pattern Recognition: The Support Vector Machine719
24.14Concluding Remarks720
24.15Bibliographical and Historical Notes722
24.16Problems723
CHAPTER 25Biologically Inspired Recognition Schemes
25.1Introduction725
25.2Artificial Neural Networks726
25.3The Backpropagation Algorithm731
25.4MLP Architectures735
25.5Overfitting to the Training Data736
25.6Optimizing the Network Architecture739
25.7Hebbian Learning740
25.8Case Study: Noise Suppression Using ANNs745
25.9Genetic Algorithms750
25.10Concluding Remarks752
25.11Bibliographical and Historical Notes753
CHAPTER 26Texture
26.1Introduction757
26.2Some Basic Approaches to Texture Analysis763
26.3Gray-level Co-occurrence Matrices764
26.4Laws’ Texture Energy Approach768
26.5Ade’s Eigenfilter Approach771
26.6Appraisal of the Laws and Ade Approaches772
26.7Fractal-based Measures of Texture774
26.8Shape from Texture775
26.9Markov Random Field Models of Texture776
26.10Structural Approaches to Texture Analysis777
26.11Concluding Remarks777
26.12Bibliographical and Historical Notes778
CHAPTER 27Image Acquisition
27.1Introduction781
27.2Illumination Schemes782
27.2.1Eliminating Shadows784
27.2.2Principles for Producing Regions of Uniform Illumination787
27.2.3Case of Two Infinite Parallel Strip Lights790
27.2.4Overview of the Uniform Illumination Scenario793
27.2.5Use of Line-scan Cameras794
27.3Cameras and Digitization796
27.3.1Digitization798
27.4The Sampling Theorem798
27.5Concluding Remarks802
27.6Bibliographical and Historical Notes803
CHAPTER 28Real-time Hardware and Systems Design Considerations
28.1Introduction805
28.2Parallel Processing806
28.3SIMD Systems807
28.4The Gain in Speed Attainable with N Processors809
28.5Flynn’s Classification810
28.6Optimal Implementation of an Image Analysis Algorithm813
28.6.1Hardware Specification and Design813
28.6.2Basic Ideas on Optimal Hardware Implementation814
28.7Some Useful Real-time Hardware Options816
28.8Systems Design Considerations818
28.9Design of Inspection Systems-The Status Quo818
28.10System Optimization822
28.11The Value of Case Studies824
28.12Concluding Remarks825
28.13Bibliographical and Historical Notes827
28.13.1General Background827
28.13.2Recent Highly Relevant Work829
PART 5PERSPECTIVES ON VISION831
CHAPTER 29Machine Vision: Art or Science?
29.1Introduction833
29.2Parameters of Importance in Machine Vision834
29.3Tradeoffs836
29.3.1Some Important Tradeoffs837
29.3.2Tradeoffs for Two-stage Template Matching838
29.4Future Directions839
29.5Hardware, Algorithms, and Processes840
29.6A Retrospective View841
29.7Just a Glimpse of Vision?842
29.8Bibliographical and Historical Notes843
APPENDIXRobust Statistics
A.1Introduction845
A.2Preliminary Definitions and Analysis848
A.3The M-estimator (Influence Function) Approach850
A.4The Least Median of Squares Approach to Regression856
A.5Overview of the Robustness Problem860
A.6The RANSAC Approach861
A.7Concluding Remarks863
A.8Bibliographical and Historical Notes864
A.9Problem865
List of Acronyms and Abbreviations867
References869
Author Index917
Subject Index925
……[看更多目录]
序言An important focus of advances in mechatronics and robotics is the addition of sensory inputs to systems with increasing "intelligence." Without doubt, sight is the "sense of choice." In everyday life, whether driving a car or threading a needle , we depend first on sight. The addition of visual perception to machines promises the greatest improvement and at the same time presents the greatest challenge.
Until relatively recently, the volume of data in the images that make up a video stream has been a serious deterrent to progress. A single frame of very modest resolution might occupy a quarter of a megabyte, so the task of handling thirty or more such frames per second requires substantial computer resources.
Fortunately, the computer and communications industries' investment in entertainment has helped address this challenge. The transmission and processing of video signals are an easy justification for selling the consumer increased computing speed and bandwidth. A digital camera, capable of video capture, has already become a fashion accessory as part of a mobile phone. As a result, video signals have become more accessible to the serious engineer. But the task of acquiring a visual image is just the tip of the iceberg.
While generating sounds and pictures is a well-defined process （speech generation is a standard "accessibility" feature of Windows）, the inverse task of recognizing connected speech is still at an unfinished state, a quarter of a century later, as any user of "dictation" software will attest. Still, analyzing sound is not even in the same league with analyzing images, particularly when they are of realworld situations rather than staged pieces with synthetic backgrounds and artificial lighting.
The task is essentially one of data reduction. From the many megabytes of the image stream, the required output might be a simple "All wheel nuts are in place" or "This tomato is ripe." But images tend to be noisy, objects that look sharp to the eye can have broken edges, boundaries can be fuzzy, and straight lines can be illusory. The task of image analysis demands a wealth of background know-how and mathematical analytic tools.
Roy Davies has been developing that rich background for well over two decades. At the time of the UK Robotics Initiative, in the 1980s, Roy had formed a relationship with the company United Biscuits. We fellow researchers might well have been amused by the task of ensuring that the blob of jam on a "Jaffacake" had been placed centrally beneath the enrobing chocolate.
文摘插图：
