Description
Feature Extraction for Image Processing and Computer Vision is an essential guide to the implementation of image processing and computer vision techniques, with tutorial introductions and sample code in MATLAB and Python. Algorithms are presented and fully explained to enable complete understanding of the methods and techniques demonstrated. As one reviewer noted, "The main strength of the proposed book is the link between theory and exemplar code of the algorithms." Essential background theory is carefully explained.This text gives students and researchers in image processing and computer vision a complete introduction to classic and state-of-the art methods in feature extraction together with practical guidance on their implementation.- The only text to concentrate on feature extraction with working implementation and worked through mathematical derivations and algorithmic methods- A thorough overview of available feature extraction methods including essential background theory, shape methods, texture and deep learning- Up to date coverage of interest point detection, feature extraction and description and image representation (including frequency domain and colour)- Good balance between providing a mathematical background and practical implementation- Detailed and explanatory of algorithms in MATLAB and Python
Table of Contents
Preface1. Introduction1.1 Overview1.2 Human and computer vision1.3 The human vision system1.3.1 The eye1.3.2 The neural system1.3.3 Processing1.4 Computer vision systems1.4.1 Cameras1.4.2 Computer interfaces1.5 Processing images1.5.1 Processing1.5.2 Hello Python, hello images!1.5.3 Mathematical tools1.5.4 Hello Matlab1.6 Associated literature1.6.1 Journals, magazines and conferences1.6.2 Textbooks1.6.3 The web1.7 ConclusionsReferences2. Images, sampling and frequency domain processing2.1 Overview2.2 Image formation2.3 The Fourier Transform2.4 The sampling criterion2.5 The discrete Fourier Transform2.5.1 One-dimensional transform2.5.2 Two-dimensional transform2.6 Properties of the Fourier Transform2.6.1 Shift invariance2.6.2 Rotation2.6.3 Frequency scaling2.6.4 Superposition (linearity)2.6.5 The importance of phase2.7 Transforms other than Fourier2.7.1 Discrete cosine transform2.7.2 Discrete Hartley Transform2.7.3 Introductory wavelets2.7.3.1 Gabor Wavelet2.7.3.2 Haar Wavelet2.7.4 Other transforms2.8 Applications using frequency domain properties2.9 Further readingReferences3. Image processing3.1 Overview3.2 Histograms3.3 Point operators3.3.1 Basic point operations3.3.2 Histogram normalisation3.3.3 Histogram equalisation3.3.4 Thresholding3.4 Group operations3.4.1 Template convolution3.4.2 Averaging operator3.4.3 On different template size3.4.4 Template convolution via the Fourier transform3.4.5 Gaussian averaging operator3.4.6 More on averaging3.5 Other image processing operators3.5.1 Median filter3.5.2 Mode filter3.5.3 Nonlocal means3.5.4 Bilateral filtering3.5.5 Anisotropic diffusion3.5.6 Comparison of smoothing operators3.5.7 Force field transform3.5.8 Image ray transform3.6 Mathematical morphology3.6.1 Morphological operators3.6.2 Grey level morphology3.6.3 Grey level erosion and dilation3.6.4 Minkowski operators3.7 Further readingReferences4. Low-level feature extraction (including edge detection)4.1 Overview4.2 Edge detection4.2.1 First-order edge detection operators4.2.1.1 Basic operators4.2.1.2 Analysis of the basic operators4.2.1.3 Prewitt edge detection operator4.2.1.4 Sobel edge detection operator4.2.1.5 The Canny edge detector4.2.2 Second-order edge detection operators4.2.2.1 Motivation4.2.2.2 Basic operators: The Laplacian4.2.2.3 The Marr–Hildreth operator4.2.3 Other edge detection operators4.2.4 Comparison of edge detection operators4.2.5 Further reading on edge detection4.3 Phase congruency4.4 Localised feature extraction4.4.1 Detecting image curvature (corner extraction)4.4.1.1 Definition of curvature4.4.1.2 Computing differences in edge direction4.4.1.3 Measuring curvature by changes in intensity (differentiation)4.4.1.4 Moravec and Harris detectors4.4.1.5 Further reading on curvature4.4.2 Feature point detection; region/patch analysis4.4.2.1 Scale invariant feature transform4.4.2.2 Speeded up robust features4.4.2.3 FAST, ORB, FREAK, LOCKY and other keypoint detectors4.4.2.4 Other techniques and performance issues4.4.3 Saliency4.4.3.1 Basic saliency4.4.3.2 Context aware saliency4.4.3.3 Other saliency operators4.5 Describing image motion4.5.1 Area-based approach4.5.2 Differential approach4.5.3 Recent developments: deep flow, epic flow and extensions4.5.4 Analysis of optical flow4.6 Further readingReferences5. High-level feature extraction: fixed shape matching5.1 Overview5.2 Thresholding and subtraction5.3 Template matching5.3.1 Definition5.3.2 Fourier transform implementation5.3.3 Discussion of template matching5.4 Feature extraction by low-level features5.4.1 Appearance-based approaches5.4.1.1 Object detection by templates5.4.1.2 Object detection by combinations of parts5.4.2 Distribution-based descriptors5.4.2.1 Description by interest points (SIFT, SURF, BRIEF)5.4.2.2 Characterising object appearance and shape5.5 Hough transform5.5.1 Overview5.5.2 Lines5.5.
