This book provides a concise introduction to computer vision for optical researchers and scientists. Building from the optical foundations of image processing and the science behind camera sensors, Optics and Artificial Vision equips the reader with the tools needed to understand and engage with digital image processing, the algorithms of optical flow and the algorithms of object detection, using Python® software to show real, implemented applications in industry. Ideal for industry engineers with projects related to computer vision, as well as a good reference text for academics, students and other researchers working at the intersection of artificial intelligence and optics.
Key Features
- Provides a concise introduction to artificial vision
- Main algorithms in computer vision are presented, step-by-step in a pedagogical way
- Presents all algorithms in Python
- Shows real implemented applications in industry
- Optical foundations of image processing are presented
Author(s): Rafael G. González-Acuña, Hector A. Chaparro-Romo, Israel Melendez-Montoya
Series: IOP Series in Emerging Technologies in Optics and Photonics
Publisher: IOP Publishing
Year: 2021
Language: English
Pages: 232
City: Bristol
PRELIMS.pdf
Preface
Acknowledgements
Acknowledgements of Rafael G González-Acuña
Acknowledgements of Héctor A Chaparro-Romo
Acknowledgements of Israel Melendez-Montoya
Author biographies
Rafael G González-Acuña
Héctor A Chaparro-Romo
Israel Melendez-Montoya
CH001.pdf
Chapter 1 Optics, sensors and images
1.1 Introduction
1.2 Optics and images
1.3 Vision
1.4 Optical instruments and optical design
1.5 Cameras
1.6 CCD sensor
1.7 CMOS sensor
1.8 Python as a program language for this book
1.9 Artificial vision and computer vision
1.10 End notes
References
CH002.pdf
Chapter 2 Introduction to computer vision
2.1 Loading and saving images
2.2 Image basics
2.3 Colour spaces
2.4 Basic image processing
2.4.1 Translation
2.4.2 Rotation
2.5 Resizing images
2.5.1 Flipping
2.5.2 Cropping
2.5.3 Image arithmetic
2.5.4 Masking
2.6 Kernels and morphological operations
2.6.1 Erosion and dilatation
2.7 Blurring
2.8 Thresholding
2.9 Gradients and edge detection
2.9.1 Gradients
2.9.2 Edges
2.10 Histograms
2.11 End notes
References
CH003.pdf
Chapter 3 Optical flow
3.1 Introduction
3.2 The Lucas–Kanade algorithm
3.2.1 Assumptions
3.2.2 The theory behind the Lucas–Kanade algorithm
3.2.3 The Lucas–Kanade algorithm step by step
3.2.4 Failures of the Lucas–Kanade algorithm
3.3 Application of the Lucas–Kanade algorithm and its Python code
3.4 The optical flow model
3.5 The Horn–Schunck algorithm
3.5.1 The smoothness principle
3.5.2 The mathematical model
3.6 End notes
References
CH004.pdf
Chapter 4 Object detection algorithms
4.1 Object detection
4.1.1 Statistical interpretation of correlation
4.1.2 Fourier interpretation of correlation
4.2 Sliding windows and image pyramids
4.3 The histogram of oriented gradients descriptor
4.4 Support vector machine
4.4.1 The concepts behind the SVM
4.5 End notes
References
CH005.pdf
Chapter 5 Image descriptors
5.1 Introduction to image descriptors
5.2 Basic statistics
5.3 Hu moments
5.4 Zernike moments
5.5 Haralick features
5.6 Local binary patterns
5.7 Keypoint detectors
5.7.1 FAST
5.7.2 The Harris method
5.7.3 GFTT
5.7.4 DoG
5.7.5 Fast Hessian
5.7.6 STAR
5.7.7 MSER
5.7.8 BRISK
5.7.9 ORB
5.8 Local invariant descriptors
5.8.1 SIFT
5.8.2 SURF
5.9 Binary descriptors
5.9.1 BRIEF
5.9.2 ORB binary descriptor
5.9.3 The BRISK binary descriptor
5.9.4 FREAK
5.10 End notes
References
CH006.pdf
Chapter 6 Neural networks
6.1 Introduction
6.2 Neural networks in a nutshell
6.3 Single perceptron learning
6.3.1 Continuous activation function perceptron
6.3.2 Single perceptron implementation
6.4 Multilayer perceptrons
6.4.1 Backpropagation
6.4.2 Maximum likelihood—binary cross-entropy
6.4.3 Maximum likelihood—multiple category cross-entropy
6.5 Convolutional neural networks
6.5.1 Introduction
6.5.2 Convolution and cross-correlation
6.5.3 Why CNNs instead of MLPs?
6.6 Metrics
6.7 CNN architectures
6.8 Transfer learning
6.9 End notes
References
CH007.pdf
Chapter 7 Optical character recognition
7.1 Introduction
7.2 Problems in classical OCR
7.3 The basic scheme of a classical OCR algorithm
7.3.1 Binarization
7.3.2 Fragmentation or segmentation of the image
7.3.3 Component thinning
7.3.4 Comparison with patterns
7.4 Classical OCR using machine learning
7.5 Modern OCR with deep learning
7.5.1 Handwritten text recognition
7.5.2 Indexing with databases
7.6 OCR with Tesseract
7.7 End notes
References
CH008.pdf
Chapter 8 Facial recognition
8.1 Introduction to facial recognition
8.2 Local binary patterns for facial recognition
8.3 The eigenfaces algorithm
8.4 Example using the CALTECH faces dataset
8.4.1 Create a personal dataset
8.5 A LBP face recognizer for your own face
8.6 Deep learning facial recognition
8.6.1 Face extraction
8.7 End notes
References
CH009.pdf
Chapter 9 Artificial vision case studies
9.1 Measuring the camera–object distance
9.1.1 Camera distortion calibration
9.1.2 Using camera sensor size or a previous distance
9.2 Single image depth estimation
9.2.1 Consistent video depth estimation
9.2.2 Adabins
9.3 State-of-the-art real-time facial detection
9.3.1 Introduction
9.4 Fruit classification
9.5 End notes
References
