Machine Learning in Image Steganalysis

Hans Georg Schaathun, Department of Computing, University of Surrey, UKDr Schaathun was previously a lecturer in coding and cryptography at the University of Bergen. Since February 2006, he has been a lecturer at the University of Surrey, UK, belonging to the research group in Digital Watermarking and Multimedia Security. His main research areas are applications of coding theory in information hiding, and machine learning techniques in steganalysis. He teaches Computer Security and Steganography at MSc level, and Functional Programming Techniques at u/g level. Dr Scaathun has published more than 35 international, peer-reviewed articles, and is an associate editor of EURASIP Journal of Information Security.

• Preface xiPART I OVERVIEW1 Introduction 31.1 Real Threat or Hype? 31.2 Artificial Intelligence and Learning 41.3 How to Read this Book 52 Steganography and Steganalysis 72.1 Cryptography versus Steganography 72.2 Steganography 82.3 Steganalysis 172.4 Summary and Notes 233 Getting Started with a Classifier 253.1 Classification 253.2 Estimation and Confidence 283.3 Using libSVM 303.4 Using Python 333.5 Images for Testing 383.6 Further Reading 39PART II FEATURES4 Histogram Analysis 434.1 Early Histogram Analysis 434.2 Notation 444.3 Additive Independent Noise 444.4 Multi-dimensional Histograms 544.5 Experiment and Comparison 635 Bit-plane Analysis 655.1 Visual Steganalysis 655.2 Autocorrelation Features 675.3 Binary Similarity Measures 695.4 Evaluation and Comparison 726 More Spatial Domain Features 756.1 The Difference Matrix 756.2 Image Quality Measures 826.3 Colour Images 866.4 Experiment and Comparison 867 The Wavelets Domain 897.1 A Visual View 897.2 The Wavelet Domain 907.3 Farid’s Features 967.4 HCF in the Wavelet Domain 987.5 Denoising and the WAM Features 1017.6 Experiment and Comparison 1068 Steganalysis in the JPEG Domain 1078.1 JPEG Compression 1078.2 Histogram Analysis 1148.3 Blockiness 1228.4 Markov Model-based Features 1248.5 Conditional Probabilities 1268.6 Experiment and Comparison 1289 Calibration Techniques 1319.1 Calibrated Features 1319.2 JPEG Calibration 1339.3 Calibration by Downsampling 1379.4 Calibration in General 1469.5 Progressive Randomisation 148PART III CLASSIFIERS10 Simulation and Evaluation 15310.1 Estimation and Simulation 15310.2 Scalar Measures 15810.3 The Receiver Operating Curve 16110.4 Experimental Methodology 17010.5 Comparison and Hypothesis Testing 17310.6 Summary 17611 Support Vector Machines 17911.1 Linear Classifiers 17911.2 The Kernel Function 18611.3 ν-SVM 18911.4 Multi-class Methods 19111.5 One-class Methods 19211.6 Summary 19612 Other Classification Algorithms 19712.1 Bayesian Classifiers 19812.2 Estimating Probability Distributions 20312.3 Multivariate Regression Analysis 20912.4 Unsupervised Learning 21212.5 Summary 21513 Feature Selection and Evaluation 21713.1 Overfitting and Underfitting 21713.2 Scalar Feature Selection 22013.3 Feature Subset Selection 22213.4 Selection Using Information Theory 22513.5 Boosting Feature Selection 23813.6 Applications in Steganalysis 23914 The Steganalysis Problem 24514.1 Different Use Cases 24514.2 Images and Training Sets 25014.3 Composite Classifier Systems 25814.4 Summary 26215 Future of the Field 26315.1 Image Forensics 26315.2 Conclusions and Notes 265Bibliography 267Index 279