An Efficient Region of Interest Encryption for HEVC/H.265 Video

다국어 초록 (Multilingual Abstract)

Efficient video encryption is essential for securing sensitive information within vast volumes of video data. Considerable research has focused on selective encryption to improve efficiency, with region of interest (ROI)-based selective encryption emerging as a critical method. The ROI-based selective encryption method targets regions containing essential visual information for encryption. Compared to methods that encrypt entire regions, the ROI-based selective encryption method enhances encryption speed and obtains contextual and situational information by encrypting only specific regions within each frame. However, ROI-based selective encryption relies on object detection algorithms to identify ROIs in each frame, which increases the overall computational complexity of the encryption process. To address this, we propose a novel ROI-based selective frame encryption technique utilizing a hierarchical B-frame structure in high-efficiency video coding (HEVC). The proposed method selects specific frames, identifies objects within them, and then encrypts them, enabling faster processing than traditional methods. We evaluate its efficiency by measuring object detection time and encryption time using detailed statistical metrics such as peak signal- to-noise ratio (PSNR), structural similarity index measure (SSIM), correlation coefficient analysis, and edge differential ratio (EDR) analysis. Additionally, to ensure that objects in all frames are fully de-identified, we introduce the Object De- Identification Rate (ODR) metric and use it to evaluate the proposed ROI encryption method. Compared to traditional methods, our approach achieves competitive ROI identification and encryption speeds while maintaining effective de-identification of ROIs in HEVC video.

목차 (Table of Contents)

• Abstract i
• List of Figures v
• List of Tables vi
• List of Abbreviations vii
• 1 Introduction 1
• 1.1 Motivation 1
• 1.2 Objectives and Contributions 3
• 1.3 Outline 6
• 2 Background 7
• 2.1 Selective Frame Encryption 9
• 2.2 Selective Encryption in CABAC 12
• 2.3 ROI-Based Selective Encryption 13
• 2.4 Security Analysis of Encrypted Video 15
• 2.5 Hierarchical B-frame Coding Structure 17
• 3 The Proposed Selective Encryption Method 22
• 3.1 Frame Selection 24
• 3.2 ROI Identification 26
• 3.3 Unit Identification and Encryption 27
• 3.4 Decryption Process of ROI Encrypted Video 29
• 4 The Proposed Quantitative Object De-Identification
• Analysis Method for Encrypted Video 31
• 4.1 Object Detection and Object Image Extraction 33
• 4.2 Object De-Identification Rate Assessment per Image 34
• 4.3 Object De-Identification Rate Assessment per Video 36
• 4.4 Security Assessment of Video De-Identification Algorithm 37
• 5 Evaluation 39
• 5.1 Performance Analysis 40
• 5.2 Security Analysis 42
• 5.2.1 PSNR and SSIM Analysis 42
• 5.2.2 Correlation Coefficient Analysis 44
• 5.2.3 Edge Differential Ratio Analysis 47
• 5.2.4 Video Differential Analysis 48
• 5.3 De-Identification Analysis 49
• 5.3.1 ODR Evaluation 50
• 5.3.2 Video Object De-Identification Rate Analysis 54
• 5.4 Experimental Results 54
• 6 Discussion 57
• 6.1 Enhancing ROI Identification Precision 57
• 6.2 Strategies to Reduce Error Propagation 58
• 6.3 Applying Robust Cryptosystems for ROI Encryption 58
• 6.4 Perceptual Anonymization and Irreversibility Evaluation 59
• 6.5 Risks of Unintended Information Leakage 60
• 7 Conclusion 61
• Reference 63
• Abstract (Korean) 71