Digital Forensics and Watermarking: 21st International Workshop, IWDW 2022, Guilin, China, November 18-19, 2022, Revised Selected Papers (Lecture Notes in Computer Science, 13825)

This book constitutes the refereed proceedings of the 21st International Workshop, IWDW 2022, held in Guilin, China, during November 18-19, 2022. The 14 full papers included in this book were carefully reviewed and selected from 30 submissions. They were organized in topical sections as follows: Steganology, Forensics and Security Analysis, Watermarking. Preface Organization Contents Steganology High-Performance Steganographic Coding Based on Sub-Polarized Channel 1 Introduction 2 Preliminaries 2.1 The Theory of Optimal Steganographic Embedding 2.2 Channel Polarization 2.3 Decoding of Polar Codes 3 Steganographic Coding on Sub-Polarized Channel 3.1 Polarized Steganographic Channel 3.2 Successive Cancellation on Polarized Steganographic Channel 3.3 Steganographic Coding Under the Typical Sub-Channel 4 Experimental Results 4.1 Construction of Polarized Steganographic Channel 4.2 Security Evaluation Under Embedding Efficiency Results 4.3 Security Evaluation Under Image Steganalysis 4.4 Evaluation of the Computational Efficiency 4.5 Discussion 5 Conclusion References High-Capacity Adaptive Steganography Based on Transform Coefficient for HEVC 1 Introduction 2 Analysis of Embedding Distortion 2.1 Embedding Distortion of Visual Degradation 2.2 Embedding Distortion of GOP 3 Proposed HEVC Steganography 3.1 Selection of Cover Coefficients 3.2 Proposed Cost Function 3.3 The Practical Implementation 4 Experimental Results 4.1 Experimental Setup 4.2 Performance of Visual Quality 4.3 Performance of Embedding Capacity 4.4 Performance of Anti-steganalysis 5 Conclusion References Forensics and Security Analysis SE-ResNet56: Robust Network Model for Deepfake Detection 1 Introduction 2 Related Works 2.1 Physiological Features Based Deepfake Detection Method 2.2 Specific Artifact Based Deepfake Detection Method 2.3 Data-Driven Deepfake Detection Method 3 The Proposed Method 3.1 Improved Residual Block SE-Res-Block 3.2 Network Framework 3.3 Deepfake Face Image Detection Algorithm Based on SE-RESNET56 4 Experimental Results and Performance Analysis 4.1 Evaluation Indicators and Dataset Settings 4.2 Ablation Experiment 4.3 The Performance Analysis 4.4 Robustness Analysis 4.5 Generalization Analysis 5 Conclusion References Voice Conversion Using Learnable Similarity-Guided Masked Autoencoder 1 Introduction 2 Related Work 3 Backgroud 3.1 Feature Decoupling for SII 4 Method 4.1 LSGM 4.2 MAE-VC 5 Experiments and Results 5.1 Experiment Conditions 5.2 Metrics 5.3 Results and Discussions 6 Conclusion References Visual Explanations for Exposing Potential Inconsistency of Deepfakes 1 Introduction 2 Related Works 2.1 Deepfake Detection 2.2 Vision Transformer 3 Method 3.1 Feature Enhancement 3.2 Multi-scale Feature Extraction 3.3 Classification 3.4 SPI 4 Experiments 4.1 Experiment Setting 4.2 Comparison Results 4.3 Distinguish Different Forgeries 4.4 Robustness to Video Compression 4.5 Visualization Analysis 5 Conclusions References Improving the Transferability of Adversarial Attacks Through Both Front and Rear Vector Method 1 Introduction 2 Related Work 2.1 Adversarial Attack Methods 2.2 Adversarial Training 3 Methodology 3.1 Previous-Gradient and Momentum as Neighborhood NI-FGSM 3.2 Both Front and Rear Vector Method 4 Experiments 4.1 Setup 4.2 Attack Single Models 4.3 Attack Ensemble Models 4.4 Ablation Study 5 Conclusion References Manipulated Face Detection and Localization Based on Semantic Segmentation 1 Introduction 2 Related Works 2.1 Face Manipulation 2.2 Manipulated Face Detection 2.3 Manipulated Face Localization 3 Methodology 3.1 Labeled Data Generation 3.2 Multi-branch Autoencoder 3.3 Loss Function 4 Experiments 4.1 Experimental Setup 4.2 Intra-dataset Evaluation 4.3 Inter-dataset Evaluation 5 Conclusion References Deep Learning Image Age Approximation - What is More Relevant: Image Content or Age Information? 1 Introduction 2 Steganalysis Residual Network (SRNet) 3 Dataset and Training Settings 3.1 PLUS Aging Dataset 3.2 Northumbria Temporal Image Forensics Database 4 Explainable Artificial Intelligence (XAI) 4.1 GradCAM++ 4.2 ScoreCAM 5 CAM Analysis 5.1 Activation on Objects 5.2 Activation on Areas 5.3 No Constant Activation Pattern 6 Potential Solutions 7 Conclusion References Watermarking Physical Anti-copying Semi-robust Random Watermarking for QR Code 1 Introduction 2 Related Work 2.1 Physical Anti-copying 2.2 Watermarking for QR Code 3 Proposed Physical Anti-copying Watermarking System 3.1 Model for Authentic and Counterfeit Channels 3.2 Watermark Embedding 3.3 Watermark Extraction and Authentication 4 Experimental Result 4.1 Experimental Setup 4.2 Comparison of the Authentic and Counterfeited QR Code 4.3 Printing Size v.s. Erroneous Bits 4.4 Printing Size v.s. Anti-copying Capability 4.5 Comparison of Authentication Performance 5 Conclusion References Robust and Imperceptible Watermarking Scheme for GWAS Data Traceability 1 Introduction 2 Genomic Data and Database Model 2.1 Genomic Data 2.2 Weighted Sum Statistic (WSS) Method 3 Proposed Database Watermarking Scheme for GWAS Data 3.1 Database Watermarking 3.2 Quantization Index Modulation (QIM) 3.3 Watermark Embedding in WSS Data 3.4 Watermark Extraction 4 Theoretical Performance 4.1 Parameter Constraints 4.2 Distortion Performance 4.3 Robustness Performance 5 Experimental Results and Discussion 5.1 Test Database 5.2 Distortion Results 5.3 Capacity Results 5.4 Robustness Results 6 Conclusion References Adaptive Robust Watermarking Method Based on Deep Neural Networks 1 Introduction 2 Proposed Method 2.1 Watermark Embedding Module 2.2 Watermark Decoding Module 2.3 Loss Function 3 Experiments 3.1 Training Process 3.2 Test Results 4 Conclusion References Adaptive Despread Spectrum-Based Image Watermarking for Fast Product Tracking 1 Introduction 2 Motivation 3 Adaptive Despread Spectrum-Based Image Watermarking 3.1 Watermark Embedding 3.2 Watermark Extraction 4 Experimental Evaluation 4.1 Experimental Setup 4.2 Robustness to Signal Processing Attacks 4.3 Robustness to Geometric Attacks 4.4 Computational Cost 5 Conclusion References Reversible Data Hiding via Arranging Blocks of Bit-Planes in Encrypted Images 1 Introduction 2 Proposed Method 2.1 Vacating Room 2.2 Image Encryption 2.3 Data Embedding 2.4 Data Extraction and Image Recovery 3 Experimental Results 3.1 Security Analysis 3.2 Embedding Performance 3.3 Performance Comparison 4 Conclusions References High Capacity Reversible Data Hiding for Encrypted 3D Mesh Models Based on Topology 1 Introduction 2 Related Works 3 Proposed Method 3.1 Room Reservation 3.2 Model Encryption 3.3 Data Embedding 3.4 Data Extraction and Model Recovery 4 Experiment Results and Analysis 4.1 Visual Quality and Quantitative Analysis 4.2 Capacity Analysis 4.3 Features Analysis 5 Conclusion References Author Index