You are previewing Machine Learning in Image Steganalysis.
O'Reilly logo
Machine Learning in Image Steganalysis

Book Description

Steganography is the art of communicating a secret message, hiding the very existence of a secret message. This is typically done by hiding the message within a non-sensitive document. Steganalysis is the art and science of detecting such hidden messages. The task in steganalysis is to take an object (communication) and classify it as either a steganogram or a clean document. Most recent solutions apply classification algorithms from machine learning and pattern recognition, which tackle problems too complex for analytical solution by teaching computers to learn from empirical data.

Part 1 of the book is an introduction to steganalysis as part of the wider trend of multimedia forensics, as well as a practical tutorial on machine learning in this context. Part 2 is a survey of a wide range of feature vectors proposed for steganalysis with performance tests and comparisons. Part 3 is an in-depth study of machine learning techniques and classifier algorithms, and presents a critical assessment of the experimental methodology and applications in steganalysis.

Key features:

  • Serves as a tutorial on the topic of steganalysis with brief introductions to much of the basic theory provided, and also presents a survey of the latest research.

  • Develops and formalises the application of machine learning in steganalysis; with much of the understanding of machine learning to be gained from this book adaptable for future study of machine learning in other applications.

  • Contains Python programs and algorithms to allow the reader to modify and reproduce outcomes discussed in the book.

  • Includes companion software available from the author's website.

Table of Contents

  1. Cover
  2. Title Page
  3. Copyright
  4. Preface
  5. Chapter 1: Introduction
    1. 1.1 Real Threat or Hype?
    2. 1.2 Artificial Intelligence and Learning
    3. 1.3 How to Read this Book
  6. Chapter 2: Steganography and Steganalysis
    1. 2.1 Cryptography versus Steganography
    2. 2.2 Steganography
    3. 2.3 Steganalysis
    4. 2.4 Summary and Notes
  7. Chapter 3: Getting Started with a Classifier
    1. 3.1 Classification
    2. 3.2 Estimation and Confidence
    3. 3.3 Using libSVM
    4. 3.4 Using Python
    5. 3.5 Images for Testing
    6. 3.6 Further Reading
  8. Chapter 4: Histogram Analysis
    1. 4.1 Early Histogram Analysis
    2. 4.2 Notation
    3. 4.3 Additive Independent Noise
    4. 4.4 Multi-dimensional Histograms
    5. 4.5 Experiment and Comparison
  9. Chapter 5: Bit-plane Analysis
    1. 5.1 Visual Steganalysis
    2. 5.2 Autocorrelation Features
    3. 5.3 Binary Similarity Measures
    4. 5.4 Evaluation and Comparison
  10. Chapter 6: More Spatial Domain Features
    1. 6.1 The Difference Matrix
    2. 6.2 Image Quality Measures
    3. 6.3 Colour Images
    4. 6.4 Experiment and Comparison
  11. Chapter 7: The Wavelets Domain
    1. 7.1 A Visual View
    2. 7.2 The Wavelet Domain
    3. 7.3 Farid's Features
    4. 7.4 HCF in the Wavelet Domain
    5. 7.5 Denoising and the WAM Features
    6. 7.6 Experiment and Comparison
  12. Chapter 8: Steganalysis in the JPEG Domain
    1. 8.1 JPEG Compression
    2. 8.2 Histogram Analysis
    3. 8.3 Blockiness
    4. 8.4 Markov Model-based Features
    5. 8.5 Conditional Probabilities
    6. 8.6 Experiment and Comparison
  13. Chapter 9: Calibration Techniques
    1. 9.1 Calibrated Features
    2. 9.2 JPEG Calibration
    3. 9.3 Calibration by Downsampling
    4. 9.4 Calibration in General
    5. 9.5 Progressive Randomisation
  14. Chapter 10: Simulation and Evaluation
    1. 10.1 Estimation and Simulation
    2. 10.2 Scalar Measures
    3. 10.3 The Receiver Operating Curve
    4. 10.4 Experimental Methodology
    5. 10.5 Comparison and Hypothesis Testing
    6. 10.6 Summary
  15. Chapter 11: Support Vector Machines
    1. 11.1 Linear Classifiers
    2. 11.2 The Kernel Function
    3. 11.3 ν-SVM
    4. 11.4 Multi-class Methods
    5. 11.5 One-class Methods
    6. 11.6 Summary
  16. Chapter 12: Other Classification Algorithms
    1. 12.1 Bayesian Classifiers
    2. 12.2 Estimating Probability Distributions
    3. 12.3 Multivariate Regression Analysis
    4. 12.4 Unsupervised Learning
    5. 12.5 Summary
  17. Chapter 13: Feature Selection and Evaluation
    1. 13.1 Overfitting and Underfitting
    2. 13.2 Scalar Feature Selection
    3. 13.3 Feature Subset Selection
    4. 13.4 Selection Using Information Theory
    5. 13.5 Boosting Feature Selection
    6. 13.6 Applications in Steganalysis
  18. Chapter 14: The Steganalysis Problem
    1. 14.1 Different Use Cases
    2. 14.2 Images and Training Sets
    3. 14.3 Composite Classifier Systems
    4. 14.4 Summary
  19. Chapter 15: Future of the Field
    1. 15.1 Image Forensics
    2. 15.2 Conclusions and Notes
  20. Bibliography
  21. Index