A Practical Introduction to Computer Vision with OpenCV, Enhanced Edition E-Book

Kenneth Dawson-Howe

• Herausgeber: John Wiley & Sons
• Kategorie: Wissenschaft und neue Technologien
• Sprache: Englisch

Seitenzahl: 342

Veröffentlichungsjahr: 2014

Beschreibung

Explains the theory behind basic computer vision and provides a bridge from the theory to practical implementation using the industry standard OpenCV libraries

Computer Vision is a rapidly expanding area and it is becoming progressively easier for developers to make use of this field due to the ready availability of high quality libraries (such as OpenCV 2).  This text is intended to facilitate the practical use of computer vision with the goal being to bridge the gap between the theory and the practical implementation of computer vision. The book will explain how to use the relevant OpenCV library routines and will be accompanied by a full working program including the code snippets from the text. This textbook is a heavily illustrated, practical introduction to an exciting field, the applications of which are becoming almost ubiquitous.  We are now surrounded by cameras, for example cameras on computers & tablets/  cameras built into our mobile phones/  cameras in games consoles; cameras imaging difficult modalities (such as ultrasound, X-ray, MRI) in hospitals, and surveillance cameras. This book is concerned with helping the next generation of computer developers to make use of all these images in order to develop systems which are more intuitive and interact with us in more intelligent ways. 

• Explains the theory behind basic computer vision and provides a bridge from the theory to practical implementation using the industry standard OpenCV libraries
• Offers an introduction to computer vision, with enough theory to make clear how the various algorithms work but with an emphasis on practical programming issues
• Provides enough material for a one semester course in computer vision at senior undergraduate and Masters levels 
• Includes the basics of cameras and images and image processing to remove noise, before moving on to topics such as image histogramming; binary imaging; video processing to detect and model moving objects; geometric operations & camera models; edge detection; features detection; recognition in images
• Contains a large number of vision application problems to provide students with the opportunity to solve real problems. Images or videos for these problems are provided in the resources associated with this book which include an enhanced eBook

Leseprobe

Contents

Cover

Title Page

Copyright Page

Dedication

List of Videos

Preface

Electronic Resources

Teaching Computer Vision Using This Text

1 Introduction

1.1 A Difficult Problem

1.2 The Human Vision System

1.3 Practical Applications of Computer Vision

1.4 The Future of Computer Vision

1.5 Material in this Textbook

1.6 Going Further with Computer Vision

2 Images

2.1 Cameras

2.2 Images

2.3 Colour Images

2.4 Noise

2.5 Smoothing

3 Histograms

3.1 1D Histograms

3.2 3D Histograms

3.3 Histogram/Image Equalisation

3.4 Histogram Comparison

3.5 Back-projection

3.6 k-means Clustering

4 Binary Vision

4.1 Thresholding

4.2 Threshold Detection Methods

4.3 Variations on Thresholding

4.4 Mathematical Morphology

4.5 Connectivity

5 Geometric Transformations

5.1 Problem Specification and Algorithm

5.2 Affine Transformations

5.3 Perspective Transformations

5.4 Specification of More Complex Transformations

5.5 Interpolation

5.6 Modelling and Removing Distortion from Cameras

6 Edges

6.1 Edge Detection

6.2 Contour Segmentation

6.3 Hough Transform

7 Features

7.1 Moravec Corner Detection

7.2 Harris Corner Detection

7.3 FAST Corner Detection

7.4 SIFT

7.5 Other Detectors

8 Recognition

8.1 Template Matching

8.2 Chamfer Matching

8.3 Statistical Pattern Recognition

8.4 Cascade of Haar Classifiers

8.5 Other Recognition Techniques

8.6 Performance

9 Video

9.1 Moving Object Detection

9.2 Tracking

9.3 Performance

10 Vision Problems

10.1 Baby Food

10.2 Labels on Glue

10.3 O-rings

10.4 Staying in Lane

10.5 Reading Notices

10.6 Mailboxes

10.7 Abandoned and Removed Object Detection

10.8 Surveillance

10.9 Traffic Lights

10.10 Real Time Face Tracking

10.11 Playing Pool

10.12 Open Windows

10.13 Modelling Doors

10.14 Determining the Time from Analogue Clocks

10.15 Which Page

10.16 Nut/Bolt/Washer Classification

10.17 Road Sign Recognition

10.18 License Plates

10.19 Counting Bicycles

10.20 Recognise Paintings

References

Index

End User License Agreement

List of Figure

Chapter 1

Figure 1.1 Different versions of an image. An array of numbers (left) which are the values of the grey scales in the low resolution image of a face (top right). The task of computer vision is most like understanding the array of numbers

Figure 1.2 PCB inspection of pads (left) and images of some detected flaws in the surface mounting of components (right). Reproduced by permission of James Mahon

Figure 1.3 Checking print quality of best-before dates (right), and monitoring level to which bottles are filled (right). Reproduced by permission of Omron Electronics LLC

Figure 1.4 Buried landmines in an infrared image (left). Reproduced by permission of Zouheir Fawaz, Handprint recognition system (right). Reproduced by permission of Siemens AG

Figure 1.5 The ASIMO humanoid robot which has two cameras in its ‘head’ which allow ASIMO to determine how far away things are, recognise familiar faces, etc. Reproduced by permission of Honda Motor Co. Inc

Chapter 2

Figure 2.1 The simple pinhole camera model showing the relationship the real 3D world (on the right-hand side) and the images captured on the image plane (on the left-hand side). The pinhole in this case is the origin in the XYZ coordinate system and, in reality, the image plane would need to be enclosed in a housing, which prevented any stray light from hitting the image plane

Figure 2.2 Four different samplings of the same image; top left 256x192, top right 128x96, bottom left 64x48 and bottom right 32x24

Figure 2.3 Four different quantizations of the same grey-scale image; top left 8 bits, top right 6 bits, bottom left 4 bits and bottom right 2 bits

Figure 2.4 RGB colour image (left) and the same image in grey-scale (right)

Figure 2.5 RGB Image (top left) shown with red channel (top right), green channel (bottom left) and blue channel (bottom right)

Figure 2.6 Spectral sensitively curves for blue (left), green (middle) and red (right) photosensitive elements

Figure 2.7 Sample arrangement of photosensitive cells in an RGB camera where the red, green and blue boxes represent individual photosensitive cells which are sensitive to wavelengths around red, green, and blue respectively. This pattern is referred as the Bayer pattern and is used in modern CCD and older CMOS cameras

Figure 2.8 CMY Image (top left) shown with yellow channel (top right), magenta channel (bottom left) and cyan channel (bottom right)

Figure 2.9 YUV image (top left) shown with luminance (Y) channel (top right), U channel (bottom left) and V channel (bottom right)

Figure 2.10 HLS space. The different colours are around the circular hue axis, the depth of the colours is indicated by how far along the saturation axis the colour is (from the centre), and the luminance indicates the brightness. The space is shown as wide in the middle and smaller at high and low values of luminance as there is no effective/reliable colour information when something is very dark or very bright

Figure 2.11 HLS Image (top left) shown with luminance channel (top right), hue channel (bottom left) and saturation channel (bottom right)

Figure 2.12 Colour image (left), selection of skin pixels by selecting all points with (Saturation >= 0.2) AND (0.5 < Luminance/Saturation <3.0) AND (Hue <= 28° OR Hue >= 330°) (right)

Figure 2.13 Selection of red-eye pixels by selecting all points with (Luminance >= 0.25) AND (Saturation >= 0.4) AND (0.5 < Luminance/Saturation <1.5) AND (Hue <= 14° OR Hue >= 324°)

Figure 2.14 Colour and grey-scale images (left) with Gaussian noise added with a mean of 0 and a standard deviation of 20 (right). The signal to noise ratios of the noisy images are 43.3 (colour image) and 40.3 (grey-scale image), with respect to the original images