Natural Image Statistics : A Probabilistic Approach to Early Computational Vision.

Aims and Scope This book is both an introductory textbook and a research monograph on modeling the statistical structure of natural images. Another motivation for natural image statistics research is in computer science and engineering, where it helps in development of better image processing and computer vision methods.

Background.- Linear Filters and Frequency Analysis.- Outline of the Visual System.- Multivariate Probability and Statistics.- Statistics of Linear Features.- Principal Components and Whitening.- Sparse Coding and Simple Cells.- Independent Component Analysis.- Information-Theoretic Interpretations.- Nonlinear Features and Dependency of Linear Features.- Energy Correlation of Linear Features and Normalization.- Energy Detectors and Complex Cells.- Energy Correlations and Topographic Organization.- Dependencies of Energy Detectors: Beyond V1.- Overcomplete and Non-negative Models.- Lateral Interactions and Feedback.- Time, Color, and Stereo.- Color and Stereo Images.- Temporal Sequences of Natural Images.- Conclusion.- Conclusion and Future Prospects.- Appendix: Supplementary Mathematical Tools.- Optimization Theory and Algorithms.- Crash Course on Linear Algebra.- The Discrete Fourier Transform.- Estimation of Non-normalized Statistical Models.