Image Processing Using Pulse-Coupled Neural Networks (eBook)

Thomas Lindblad is a professor at the Royal Institute of Technology (Physics) in Stockholm. Working and teaching nuclear and environmental physics his main interest is with sensors, signal processing and intelligent data analysis of torrent data from experiments on-line accelerators, in space, etc. Jason Kinser is an associate professor at George Mason University. He has developed a plethora of image processing applications in the medical, military, and industrial fields. He has been responsible for the conversion of PCNN theory into practical applications providing many improvements in both speed and performance.

Preface 5
Preface to the First Edition 7
Contents 9
1 Introduction and Theory 12
1.1 General Aspects 12
1.2 The State of Traditional Image Processing 13
1.2.1 Generalisation versus Discrimination 13
1.2.2 The World of Inner Products” 14
1.2.3 The Mammalian Visual System 15
1.2.4 Where Do We Go From Here? 15
1.3 Visual Cortex Theory 16
1.3.1 A Brief Overview of the Visual Cortex 16
1.3.2 The Hodgkin–Huxley Model 17
1.3.3 The Fitzhugh–Nagumo Model 18
1.3.4 The Eckhorn Model 19
1.3.5 The Rybak Model 20
1.3.6 The Parodi Model 21
1.4 Summary 21
2 Theory of Digital Simulation 22
2.1 The Pulse-Coupled Neural Network 22
2.1.1 The Original PCNN Model 22
2.1.2 Time Signatures 27
2.1.3 The Neural Connections 29
2.1.4 Fast Linking 32
2.1.5 Fast Smoothing 33
2.1.6 Analogue Time Simulation 34
2.2 The ICM – A Generalized Digital Model 35
2.2.1 Minimum Requirements 36
2.2.2 The ICM 37
2.2.3 Interference 38
2.2.4 Curvature Flow Models 42
2.2.5 Centripetal Autowaves 43
2.3 Summary 45
3 Automated Image Object Recognition 46
3.1 Important Image Features 46
3.2 Image Segmentation – A Red Blood Cell Example 52
3.3 Image Segmentation – A Mammography Example 53
3.4 Image Recognition – An Aircraft Example 54
3.5 Image Classi.cation – Aurora Borealis Example 55
3.6 The Fractional Power Filter 57
3.7 Target Recognition – Binary Correlations 58
3.8 Image Factorisation 62
3.9 A Feedback Pulse Image Generator 63
3.10 Object Isolation 66
3.11 Dynamic Object Isolation 69
3.12 Shadowed Objects 71
3.13 Consideration of Noisy Images 73
3.14 Summary 78
4 Image Fusion 80
4.1 The Multi-spectral Model 80
4.2 Pulse-Coupled Image Fusion Design 82
4.3 A Colour Image Example 84
4.4 Example of Fusing Wavelet Filtered Images 86
4.5 Detection of Multi-spectral Targets 86
4.6 Example of Fusing Wavelet Filtered Images 91
4.7 Summary 92
5 Image Texture Processing 94
5.1 Pulse Spectra 94
5.2 Statistical Separation of the Spectra 98
5.3 Recognition Using Statistical Methods 99
5.4 Recognition of the Pulse Spectra via an Associative Memory 100
5.5 Summary 103
6 Image Signatures 104
6.1 Image Signature Theory 104
6.1.1 The PCNN and Image Signatures 105
6.1.2 Colour Versus Shape 106
6.2 The Signatures of Objects 106
6.3 The Signatures of Real Images 108
6.4 Image Signature Database 110
6.5 Computing the Optimal Viewing Angle 111
Nils Zetterlund 111
6.6 Motion Estimation 114
6.7 Summary 117
7 Miscellaneous Applications 118
7.1 Foveation 118
7.1.1 The Foveation Algorithm 119
7.1.2 Target Recognition by a PCNN Based Foveation Model 121
7.2 Histogram Driven Alterations 124
7.3 Maze Solutions 126
7.4 Barcode Applications 127
Soonil D.D.V. Rughooputh and Harry C.S. Rughooputh 127
7.4.1 Barcode Generation from Data Sequence and Images 128
Binary Code for a Still Image 128
Algorithm 128
Binary Barcode for Data Sequences 130
Algorithm 130
7.4.2 PCNN Counter 132
7.4.3 Chemical Indexing 132
Current Searching Technique 135
7.4.4 Identi.cation and Classi.cation of Galaxies 137
7.4.5 Navigational Systems 142
7.4.6 Hand Gesture Recognition 145
7.4.7 Road Surface Inspection 148
7.5 Summary 152
8 Hardware Implementations 154
8.1 Theory of Hardware Implementation 154
8.2 Implementation on a CNAPs Processor 155
8.3 Implementation in VLSI 157
8.4 Implementation in FPGA 157
8.5 An Optical Implementation 162
8.6 Summary 164
References 166
Index 174