Computational Color Science

Edoardo Provenzi is Associate Professor at the Paris Descartes University in France. His main research interest is theoretical and applied color image science.

• Preface ixChapter 1. Rudiments of Human Visual System (HVS) Features 11.1. The retina 11.1.1. Photoreceptors: rods and cones 21.2. Adaptation and photo-electrical response of receptors 41.3. Spatial locality of vision 51.4. Local contrast enhancement 61.5. Physical vs. perceived light intensity contrast: Weber- Fechner’s law 9Chapter 2. Computational Color Constancy Algorithms 132.1. The dichromatic and Lambertian image formation models 142.2. Classical hypotheses for illuminant and reflectance estimation 172.2.1. White-patch assumption and related models 182.2.2. Gray-world assumption and related models 202.2.3. Shades of gray and multi-scale max-RGB assumptions to mix white-patch and gray-world hypotheses 222.2.4. Gray-edge assumption and related models 242.2.5. Multi-scale n-th order shades of gray-edge assumption: a general hypothesis 26Chapter 3. Retinex-like Algorithms for Color Image Processing 293.1. Mathematical description of the original ratio-threshold-reset Retinex algorithm 303.2. Analysis of the ratio-reset Retinex formula: the limit ε → 0 333.2.1. Retinex: “a melody that everyone plays differently” 373.3. From paths to pixel sprays: RSR 413.3.1. LRSR and SMRSR 433.4. A psychophysical method to measure (achromatic) induction 453.5. Automatic Color Equalization: ACE 503.6. RACE: a model with mixed features between RSR and ACE 523.6.1. Regularization of RACE formula: attachment to original image 543.7. An alternative fusion between RSR and ACE: STRESS 56Chapter 4. Variational Formulation of Histogram Equalization 594.1. The Caselles–Sapiro model 594.2. Interpretation of Caselles–Sapiro’s functional for histogram equalization 654.3. Application of histogram equalization techniques to color images 67Chapter 5. Perceptually-inspired Variational Models for Color Enhancement in the RGB Space 695.1. Beyond the Caselles–Sapiro model: modification of the histogram equalization functional to approach visual properties 705.1.1. A contrast term coherent with HVS properties 705.1.2. Entropic adjustment term 735.2. Minimization of perceptual functionals 755.2.1. Stability of the iterative semi-implicit gradient descent scheme 805.2.2. A general strategy for the reduction of computational complexity 815.2.3. Results 835.3. Embedding existing perceptually inspired color correction models in the variational framework 855.3.1. Alternative variational and EDP formalizations of Retinex-like algorithms 895.4. Variational interpretation of the Rudd-Zemach model of achromatic induction 925.5. Perceptual enhancement in the wavelet domain 955.5.1. Adjustment to the average value in the wavelet domain 985.5.2. Local contrast enhancement in the wavelet domain 995.6. High-dynamic-range (HDR) imaging 1015.6.1. A two-stage tone mapping 105Appendix 111Bibliography 117Index 125