Color in Computer Vision -- Contents -- Preface -- 1 Introduction -- 1.1 From Fundamental to Applied -- 1.2 Part I: Color Fundamentals -- 1.3 Part II: Photometric Invariance -- 1.3.1 Invariance Based on Physical Properties -- 1.3.2 Invariance By Machine Learning -- 1.4 Part III: Color Constancy -- 1.5 Part IV: Color Feature Extraction -- 1.5.1 From Luminance to Color -- 1.5.2 Features, Descriptors, and Saliency -- 1.5.3 Segmentation -- 1.6 Part V: Applications -- 1.6.1 Retrieval and Visual Exploration -- 1.6.2 Color Naming -- 1.6.3 Multispectral Applications -- 1.7 Summary -- PART I Color Fundamentals -- 2 Color Vision -- 2.1 Introduction -- 2.2 Stages of Color Information Processing -- 2.2.1 Eye and Optics -- 2.2.2 Retina: Rods and Cones -- 2.2.3 Ganglion Cells and Receptive Fields -- 2.2.4 LGN and Visual Cortex -- 2.3 Chromatic Properties of the Visual System -- 2.3.1 Chromatic Adaptation -- 2.3.2 Human Color Constancy -- 2.3.3 Spatial Interactions -- 2.3.4 Chromatic Discrimination and Color Deficiency -- 2.4 Summary -- 3 Color Image Formation -- 3.1 Lambertian Reflection Model -- 3.2 Dichromatic Reflection Model -- 3.3 Kubelka-Munk Model -- 3.4 The Diagonal Model -- 3.5 Color Spaces -- 3.5.1 XYZ System -- 3.5.2 RGB System -- 3.5.3 Opponent Color Spaces -- 3.5.4 Perceptually Uniform Color Spaces -- 3.5.5 Intuitive Color Spaces -- 3.6 Summary -- PART II Photometric Invariance -- 4 Pixel-Based Photometric Invariance -- 4.1 Normalized Color Spaces -- 4.2 Opponent Color Spaces -- 4.3 The HSV Color Space -- 4.4 Composed Color Spaces -- 4.4.1 Body Reflectance Invariance -- 4.4.2 Body and Surface Reflectance Invariance -- 4.5 Noise Stability and Histogram Construction -- 4.5.1 Noise Propagation -- 4.5.2 Examples of Noise Propagation through Transformed Colors -- 4.5.3 Histogram Construction by Variable Kernel Density Estimation.

4.6 Application: Color-Based Object Recognition -- 4.6.1 Dataset and Performance Measure -- 4.6.2 Robustness Against Noise: Simulated Data -- 4.7 Summary -- 5 Photometric Invariance from Color Ratios -- 5.1 Illuminant Invariant Color Ratios -- 5.2 Illuminant Invariant Edge Detection -- 5.3 Blur-Robust and Color Constant Image Description -- 5.4 Application: Image Retrieval Based on Color Ratios -- 5.4.1 Robustness to Illuminant Color -- 5.4.2 Robustness to Gaussian Blur -- 5.4.3 Robustness to Real-World Blurring Effects -- 5.5 Summary -- 6 Derivative-Based Photometric Invariance -- 6.1 Full Photometric Invariants -- 6.1.1 The Gaussian Color Model -- 6.1.2 The Gaussian Color Model by an RGB Camera -- 6.1.3 Derivatives in the Gaussian Color Model -- 6.1.4 Differential Invariants for the Lambertian Reflection Model -- 6.1.5 Differential Invariants for the Dichromatic Reflection Model -- 6.1.6 Summary of Full Color Invariants -- 6.1.7 Geometrical Color Invariants in Two Dimensions -- 6.2 Quasi-Invariants -- 6.2.1 Edges in the Dichromatic Reflection Model -- 6.2.2 Photometric Variants and Quasi-Invariants -- 6.2.3 Relations of Quasi-Invariants with Full Invariants -- 6.2.4 Localization and Discriminative Power of Full and Quasi-Invariants -- 6.3 Summary -- 7 Photometric Invariance by Machine Learning -- 7.1 Learning from Diversified Ensembles -- 7.2 Temporal Ensemble Learning -- 7.3 Learning Color Invariants for Region Detection -- 7.4 Experiments -- 7.4.1 Error Measures -- 7.4.2 Skin Detection: Still Images -- 7.4.3 Road Detection in Video Sequences -- 7.5 Summary -- PART III Color Constancy -- 8 Illuminant Estimation and Chromatic Adaptation -- 8.1 Illuminant Estimation -- 8.2 Chromatic Adaptation -- 9 Color Constancy Using Low-level Features -- 9.1 General Gray-World -- 9.2 Gray-Edge -- 9.3 Physics-Based Methods -- 9.4 Summary.

10 Color Constancy Using Gamut-Based Methods -- 10.1 Gamut Mapping Using Derivative Structures -- 10.1.1 Diagonal-Offset Model -- 10.1.2 Gamut Mapping of Linear Combinations of Pixel Values -- 10.1.3 N-Jet Gamuts -- 10.2 Combination of Gamut Mapping Algorithms -- 10.2.1 Combining Feasible Sets -- 10.2.2 Combining Algorithm Outputs -- 10.3 Summary -- 11 Color Constancy Using Machine Learning -- 11.1 Probabilistic Approaches -- 11.2 Combination Using Output Statistics -- 11.3 Combination Using Natural Image Statistics -- 11.3.1 Spatial Image Structures -- 11.3.2 Algorithm Selection -- 11.4 Methods Using Semantic Information -- 11.4.1 Using Scene Categories -- 11.4.2 Using High-Level Visual Information -- 11.5 Summary -- 12 Evaluation of Color Constancy Methods -- 12.1 Data Sets -- 12.1.1 Hyperspectral Data -- 12.1.2 RGB Data -- 12.1.3 Summary -- 12.2 Performance Measures -- 12.2.1 Mathematical Distances -- 12.2.2 Perceptual Distances -- 12.2.3 Color Constancy Distances -- 12.2.4 Perceptual Analysis -- 12.3 Experiments -- 12.3.1 Comparing Algorithm Performance -- 12.3.2 Evaluation -- 12.4 Summary -- PART IV Color Feature Extraction -- 13 Color Feature Detection -- 13.1 The Color Tensor -- 13.1.1 Photometric Invariant Derivatives -- 13.1.2 Invariance to Color Coordinate Transformations -- 13.1.3 Robust Full Photometric Invariance -- 13.1.4 Color-Tensor-Based Features -- 13.1.5 Experiment: Robust Feature Point Detection and Extraction -- 13.2 Color Saliency -- 13.2.1 Color Distinctiveness -- 13.2.2 Physics-Based Decorrelation -- 13.2.3 Statistics of Color Images -- 13.2.4 Boosting Color Saliency -- 13.2.5 Evaluation of Color Distinctiveness -- 13.2.6 Repeatability -- 13.2.7 Illustrations of Generality -- 13.3 Conclusions -- 14 Color Feature Description -- 14.1 Gaussian Derivative-Based Descriptors -- 14.2 Discriminative Power -- 14.3 Level of Invariance.

14.4 Information Content -- 14.4.1 Experimental Results -- 14.5 Summary -- 15 Color Image Segmentation -- 15.1 Color Gabor Filtering -- 15.2 Invariant Gabor Filters Under Lambertian Reflection -- 15.3 Color-Based Texture Segmentation -- 15.4 Material Recognition Using Invariant Anisotropic Filtering -- 15.4.1 MR8-NC Filterbank -- 15.4.2 MR8-INC Filterbank -- 15.4.3 MR8-LINC Filterbank -- 15.4.4 MR8-SLINC Filterbank -- 15.4.5 Summary of Filterbank Properties -- 15.5 Color Invariant Codebooks and Material-Specific Adaptation -- 15.6 Experiments -- 15.6.1 Material Classification by Color Invariant Codebooks -- 15.6.2 Color-Texture Segmentation of Material Images -- 15.6.3 Material Classification by Adaptive Color Invariant Codebooks -- 15.7 Image Segmentation by Delaunay Triangulation -- 15.7.1 Homogeneity Based on Photometric Color Invariance -- 15.7.2 Homogeneity Based on a Similarity Predicate -- 15.7.3 Difference Measure -- 15.7.4 Segmentation Results -- 15.8 Summary -- PART V Applications -- 16 Object and Scene Recognition -- 16.1 Diagonal Model -- 16.2 Color SIFT Descriptors -- 16.3 Object and Scene Recognition -- 16.3.1 Feature Extraction Pipelines -- 16.3.2 Classification -- 16.3.3 Image Benchmark: PASCAL Visual Object Classes Challenge -- 16.3.4 Video Benchmark: Mediamill Challenge -- 16.3.5 Evaluation Criteria -- 16.4 Results -- 16.4.1 Image Benchmark: PASCAL VOC Challenge -- 16.4.2 Video Benchmark: Mediamill Challenge -- 16.4.3 Comparison -- 16.5 Summary -- 17 Color Naming -- 17.1 Basic Color Terms -- 17.2 Color Names from Calibrated Data -- 17.2.1 Fuzzy Color Naming -- 17.2.2 Chromatic Categories -- 17.2.3 Achromatic Categories -- 17.2.4 Fuzzy Sets Estimation -- 17.3 Color Names from Uncalibrated Data -- 17.3.1 Color Name Data Sets -- 17.3.2 Learning Color Names -- 17.3.3 Assigning Color Names in Test Images.

17.3.4 Flexibility Color Name Data Set -- 17.4 Experimental Results -- 17.5 Conclusions -- 18 Segmentation of Multispectral Images -- 18.1 Reflection and Camera Models -- 18.1.1 Multispectral Imaging -- 18.1.2 Camera and Image Formation Models -- 18.1.3 White Balancing -- 18.2 Photometric Invariant Distance Measures -- 18.2.1 Distance between Chromaticity Polar Angles -- 18.2.2 Distance between Hue Polar Angles -- 18.2.3 Discussion -- 18.3 Error Propagation -- 18.3.1 Propagation of Uncertainties due to Photon Noise -- 18.3.2 Propagation of Uncertainty -- 18.4 Photometric Invariant Region Detection by Clustering -- 18.4.1 Robust K-Means Clustering -- 18.4.2 Photometric Invariant Segmentation -- 18.5 Experiments -- 18.5.1 Propagation of Uncertainties in Transformed Spectra -- 18.5.2 Photometric Invariant Clustering -- 18.6 Summary -- Citation Guidelines -- References -- Index.