Deep Learning for Multimedia Processing Applications: Volume 1: Image Security and Intelligent Systems for Multimedia Processing

✍ Scribed by Uzair Aslam Bhatti, Huang Mengxing, Jingbing Li, Sibghat Ullah Bazai, Muhammad Aamir Copyright 2024

• Publisher: CRC Press
• Year: 2023
• Tongue: English
• Leaves: 313
• Series: Deep Learning for Multimedia Processing Applications
• Category: Library

✦ Synopsis

Deep Learning for Multimedia Processing Applications is a comprehensive guide that explores the revolutionary impact of deep learning techniques in the field of multimedia processing. Written for a wide range of readers, from students to professionals, this book offers a concise and accessible overview of the application of deep learning in various multimedia domains, including image processing, video analysis, audio recognition, and natural language processing.

Divided into two volumes, Volume One begins by introducing the fundamental concepts of deep learning, providing readers with a solid foundation to understand its relevance in multimedia processing. Readers will discover how deep learning techniques enable accurate and efficient image recognition, object detection, semantic segmentation, and image synthesis. The book also covers video analysis techniques, including action recognition, video captioning, and video generation, highlighting the role of deep learning in extracting meaningful information from videos.

Furthermore, the book explores audio processing tasks such as speech recognition, music classification, and sound event detection using deep learning models. It demonstrates how deep learning algorithms can effectively process audio data, opening up new possibilities in multimedia applications. Lastly, the book explores the integration of deep learning with natural language processing techniques, enabling systems to understand, generate, and interpret textual information in multimedia contexts.

Throughout the book, practical examples, code snippets, and real-world case studies are provided to help readers gain hands-on experience in implementing deep learning solutions for multimedia processing. Deep Learning for Multimedia Processing Applications is an essential resource for anyone interested in harnessing the power of deep learning to unlock the vast potential of multimedia data.

✦ Table of Contents

Cover
Half Title
Title Page
Copyright Page
Contents
Chapter 1. A Novel Robust Watermarking Algorithm for Encrypted Medical Images Based on Non-Subsampled Shearlet Transform and Schur Decomposition
1.1 Introduction
1.2 Basic Theory
1.2.1 Discrete Wavelet Transform (DWT)
1.2.2 Non-Subsampled Shearlet Transform (NSST)
1.2.3 Matrix Schur Decomposition
1.2.4 Chaos Encryption System
1.2.4.1 Tent
1.2.4.2 Logistic Map
1.3 Proposed Algorithm
1.3.1 Medical Image Encryption
1.3.2 Feature Extraction
1.3.3 Embed Watermark
1.3.4 Extraction of Watermark
1.4 Experiments and Analysis of Results
1.4.1 Simulation Experiment
1.4.2 Attacks Results
1.4.3 Contrastion to Plaintext Domain Algorithm
1.4.4 Contrastion to Other Encrypted Algorithms
1.5 Conclusion
References
Chapter 2. Robust Zero Watermarking Algorithm for Encrypted Medical Images Based on SUSAN-DCT
2.1 Introduction
2.2 Literature Review
2.3 Basic Theory and Proposed Algorithm
2.3.1 SUSAN Edge Detection
2.3.2 Hu Moments
2.3.3 Logical Mapping
2.3.4 Proposed Algorithm
2.3.4.1 Medical Image Encryption
2.3.4.2 Watermark Encryption
2.3.4.3 Watermark Embedding
2.3.4.4 Extraction and Decryption of Watermark
2.4 Experiment and Results
2.4.1 Evaluation Parameter
2.4.2 Experimental Setup
2.4.3 Results and Analysis
2.4.3.1 Gaussian Noise Attack
2.4.3.2 JPEG Compression Attack
2.4.3.3 Median Filter Attack
2.4.3.4 Rotation Attack
2.4.3.5 Scaling Attack
2.4.3.6 Translation Attack
2.4.3.7 Cropping Attack
2.5 Conclusion
References
Chapter 3. Robust Zero Watermarking Algorithm for Encrypted Medical Volume Data Based on PJFM and 3D-DCT
3.1 Introduction
3.2 The Fundamental Theory
3.2.1 Pseudo Jacobi-Fourier Moment
3.2.2 D-DCT and 3D-IDCT
3.2.3 Logistic Mapping
3.3 The Proposed Method
3.3.1 Medical Volume Data Encryption
3.3.2 Feature Extraction
3.3.3 Watermark Encryption and Embedding
3.3.4 Watermark Extraction and Decryption
3.4 Experimental Results and Performance Evaluation
3.4.1 Simulation Experiment
3.4.2 Attacks Results
3.4.3 Comparison with Unencrypted Algorithm
3.5 Conclusion
References
Chapter 4. Robust Zero Watermarking Algorithm for Medical Images Based on BRISK and DCT
4.1 Introduction
4.2 Fundamental Theory
4.2.1 BRISK Feature Extraction Algorithm
4.2.1.1 Scale Space Keypoints Detection
4.2.1.2 Keypoints Description
4.2.2 Discrete Cosine Transform (DCT)
4.2.3 Logistic Mapping
4.3 Proposed Algorithm
4.3.1 Medical Image Feature Extraction
4.3.2 Watermark Encryption
4.3.3 Embed Watermark
4.3.4 Watermark Extraction and Decryption
4.4 Experiments and Results
4.4.1 Test Different Images
4.4.2 Conventional Attacks
4.4.3 Geometric Attacks
4.4.4 Compare with Other Algorithms
4.5 Conclusion
References
Chapter 5. Robust Color Images Zero-Watermarking Algorithm Based on Stationary Wavelet Transform and Daisy Descriptor
5.1 Introduction
5.2 Literature Review
5.3 Material and Techniques
5.3.1 Daisy Descriptor
5.3.2 Stationary Wavelet Transform
5.3.3 Tent Chaotic Map
5.3.4 Proposed Algorithm
5.3.4.1 Feature Extraction
5.3.4.2 Watermark Encryption and Embedding
5.3.4.3 Watermark Extraction and Recovery
5.4 Experiment and Results
5.4.1 Evaluation Parameter
5.4.2 Feasibility Analysis
5.4.2.1 Distinguishability Analysis
5.4.2.2 Robustness Analysis
5.4.3 Results and Analysis
5.4.3.1 Conventional Attack
5.4.3.2 Geometric Attack
5.5 Conclusion
References
Chapter 6. Robust Multi-watermarking Algorithm Based on DarkNet53
6.1 Introduction
6.2 Basic Theory
6.2.1 DarkNet53
6.2.2 Discrete Cosine Transform
6.2.3 Logistic Map
6.3 Proposed Algorithm
6.3.1 Improvement of DarkNet53 Network Model
6.3.1.1 Improvement of Network Structure
6.3.1.2 Data Set Creation
6.3.1.3 Training Network
6.3.2 Encryption of Watermark
6.3.3 Watermark Embedding
6.3.4 Extraction of a Watermark
6.3.5 Decryption of a Watermark
6.4 Experimental Results and Analysis
6.4.1 Performance
6.4.2 Reliability Analysis
6.4.3 Traditional Attack
6.4.4 Geometric Attack
6.5 Conclusion
References
Chapter 7. Robust Multi-watermark Algorithm for Medical Images Based on SqueezeNet Transfer Learning
7.1 Introduction
7.2 Fundamental Theory
7.2.1 SqueezeNet Neural Network
7.2.2 Transfer Learning
7.2.3 SPM Composite Chaotic Mapping
7.3 Proposed Algorithm
7.3.1 Retraining the Network
7.3.2 Watermark Encryption
7.3.3 Generation and Extraction of Zero Watermark
7.3.4 Decryption of Watermark
7.4 Experimental Results
7.4.1 Evaluation Metrics
7.4.2 Discrimination Testing
7.4.3 Robustness Testing
7.4.4 Comparison
7.5 Conclusion
References
Chapter 8. Deep Learning Applications in Digital Image Security: Latest Methods and Techniques
8.1 Introduction
8.2 Background
8.2.1 Basic Model
8.2.2 Learning-based Model
8.3 Classification of Digital Watermarking
8.3.1 Divided by Characteristics
8.3.1.1 Robust Watermarking
8.3.1.2 Fragile Watermark
8.3.2 Divided by Detection Method
8.3.2.1 Blind Watermark
8.3.2.2 Non-blind Watermarking
8.3.2.3 Zero Watermark
8.3.3 Divided by Hidden Domain
8.3.3.1 Watermarking Algorithm Based on Frequency Domain
8.3.3.2 Watermarking Algorithm Based on Spatial Domain
8.3.4 Other Classifications
8.4 Performance Evaluation and Algorithms
8.4.1 Performance Evaluation
8.4.2 Algorithms
8.5 Attacks
8.5.1 Robust Attack
8.5.1.1 Non-geometric Attacks
8.5.1.2 Geometric Attack
8.5.1.3 Combination Attack
8.5.1.4 Removal Attack
8.5.1.5 Protocol Attack
8.5.1.6 Copy Attack
8.5.1.7 Cryptographic Attack
8.5.2 No Attack
8.5.3 Explaining the Attack
8.6 Learning-based Watermarking
8.7 Applications of Learning-based Watermarking
8.7.1 Medical Field
8.7.2 Remote-sensing Field
8.7.3 Map Copyright
8.7.4 Copyright Protection
8.7.5 Content Authentication
8.7.6 Infringement Tracking
8.7.7 Radio Monitoring
8.7.8 Copy Control
8.7.9 Electronic Field
8.8 Conclusion
Funding
References
Chapter 9. Image Fusion Techniques and Applications for Remote Sensing and Medical Images
9.1 Introduction
9.2 Rule of Image Fusion
9.3 Levels of Image Fusion
9.3.1 Pixel-Level Image Fusion
9.3.2 Feature-Level Image Fusion
9.3.3 Decision-Level Image Fusion
9.4 Image Fusion Methods
9.4.1 Spatial Domain Fusion Methods
9.4.1.1 Intensity Hue Saturation
9.4.1.2 Brovey Transform
9.4.1.3 Principal Component Analysis (PCA)
9.4.1.4 High-Pass Filtering
9.4.2 Frequency Domain Fusion Methods
9.4.2.1 Laplacian Pyramid Fusion Technique
9.4.2.2 Discrete Cosine Transform
9.4.2.3 Wavelet Transform
9.4.2.4 Kekre's Wavelet Transform
9.4.2.5 Kekre's Hybrid Wavelet Transform
9.4.2.6 Stationary Wavelet Transform
9.4.2.7 Curvelet Transform
9.4.3 Deep Learning Methods
9.5 Techniques for the Assessment of Image Fusion Quality
9.6 Image Fusion Categorization
9.6.1 Single Sensor
9.6.2 Multi-Sensors
9.6.3 Multiview Fusion
9.6.4 Multimodal Fusion
9.6.5 Multi-Focus Fusion
9.6.6 Multi-Temporal Fusion
9.7 Image Fusion Applications
9.7.1 Medical Image Fusion
9.7.1.1 Types of Medical Images
9.7.1.2 Combining of Imaging Modalities
9.7.2 Remote-Sensing Image Fusion
9.7.2.1 Types of Satellite Images
9.7.2.1.1 Single-Band Images
9.7.2.1.2 Multispectral Images
9.7.3 Visible-Infrared Fusion
9.7.3.1 Types of Visible-Infrared Images
9.7.3.1.1 False-Color Composites
9.7.3.1.2 Grayscale Composites
9.7.3.1.3 Merged Images
9.7.4 Multi-Focus Image Fusion
9.8 Conclusion
References
Chapter 10. Detecting Phishing URLs through Deep Learning Models
10.1 Introduction
10.2 DL Models Used in Cybersecurity
10.2.1 Convolutional Neural Network
10.2.2 Recurrent Neural Networks
10.2.3 Long Short-Term Memory
10.2.4 Deep Belief Networks
10.2.5 Multi-Layer Perceptron
10.2.6 Generative Adversarial Network
10.3 Metrics
10.3.1 Accuracy
10.3.2 Precision
10.3.3 Recall (Sensitivity)
10.3.4 F1 Score
10.3.5 Confusion Matrix
10.4 Application of Deep Learning in Cybersecurity Use Cases
10.4.1 Intrusion Detection System
10.4.2 Malware Detection
10.4.3 Botnet Detection
10.4.4 Network Traffic Identification
10.4.5 Credit Card Fraud Detection
10.5 Existing Work Related to Phishing URL Detection Using DL Models
10.6 Conclusion
References
Chapter 11. Augmenting Multimedia Analysis: A Fusion of Deep Learning with Differential Privacy
11.1 Introduction
11.2 Multimedia Data and Crowdsensing Privacy Concerns
11.3 Deep Learning and Privacy Risks
11.3.1 Privacy Attacks in Deep Learning Pipeline
11.4 Algorithms for Preserving Privacy
11.5 The Differential Privacy Distributions
11.6 How Differential Privacy Fuses With Deep Learning
11.7 Methodology: Exploring the Intersection of Multimedia Data With Deep Learning and Privacy in Literature
11.7.1 Preserving-Privacy Image Analysis
11.7.2 Preserving-Privacy Video Analysis
11.7.3 Preserving-Privacy With Other Methods
11.8 Discussion
11.9 Conclusion
References
Chapter 12. Multi-classification Deep Learning Models for Detecting Multiple Chest Infection Using Cough and Breath Sounds
12.1 Introduction
12.2 Literature Review
12.3 Materials and Methods
12.3.1 Proposed Study Flow for the Diagnosis of Multiple Chest Infections
12.3.2 Data Set Description
12.3.3 Using SMOTE Tomek to Balance the Data Set
12.3.4 Deep Learning Classifiers
12.3.5 Proposed Model
12.3.5.1 Structure of the Proposed DMCIC_Net
12.3.5.2 Convolutional Blocks of CNN Model
12.3.5.3 Flattened Layer
12.3.5.4 Dropout Layer
12.3.6 Dense Block of Proposed Model
12.3.6.1 ReLU Activation
12.3.6.2 Dense Layer
12.3.7 Model Evaluations
12.3.7.1 Accuracy
12.3.7.2 Precision
12.3.7.3 Recall
12.3.7.4 F1-Score
12.4 Results and Discussion
12.4.1 Experimental Setup
12.4.2 Accuracy Comparison of Proposed Model with Baseline Models
12.4.3 AUC Comparison with Baseline Models
12.4.4 Comparison with Baseline Models Using Precision
12.4.5 Comparison of DMCIC_Net with Baseline Models Using Recall
12.4.6 F1-Score Comparison with Baseline Models
12.4.7 Comparison of Proposed Model with Baseline Models Using Loss
12.4.8 Comparison of ROC with Current Models
12.4.9 AU(ROC) Extension for Multiclass Comparison Against Recent Models
12.4.10 Comparison of DMCIC_Net with Six Models Using a Confusion Matrix
12.4.11 Comparison of the Proposed Model with State of the Art
12.4.12 Discussion
12.5 Conclusion
References
Chapter 13. Classifying Traffic Signs Using Convolutional Neural Networks Based on Deep Learning Models
13.1 Introduction
13.2 How Does a Model Learn?
13.2.1 Types of Machine Learning
13.2.2 Tasks Performed by Machine Learning
13.2.3 Depth of Machine Learning
13.3 Deep Learning
13.3.1 Training of Deep Learning Models
13.3.2 Algorithms Used to Train Deep Learning Models
13.4 Classification of Images Using a Convolutional Neural Network
13.4.1 Classifying Images Using Traditional Methods
13.4.2 Image Classification Using CNN
13.4.3 Overview of CNN Models Used for Image Classification
13.5 Classifying Traffic Signs Using Convolutional Neural Network
13.5.1 Model Architecture
13.5.1.1 Input Layer
13.5.1.2 Convolutional Layer
13.5.1.3 Activation Function
13.5.1.4 Pooling Layers
13.5.1.5 Dropout Layer
13.5.1.6 Flatten Layer
13.5.1.7 Fully Connected Layer
13.5.1.8 Output Layer
13.5.1.9 Loss Function
13.5.1.10 Training and Testing
13.5.2 Benchmark on Traffic Sign Classification
13.6 Data Set
13.6.1 German Traffic Sign Recognition Benchmark
13.6.2 Comparison With Other Data Sets
13.7 Experimental Setup
13.8 Conclusion
References
Chapter 14. Cloud-Based Intrusion Detection System Using a Deep Neural Network and Human-in-the-Loop Decision Making
14.1 Introduction
14.2 Related Work
14.3 Preliminaries
14.3.1 Cloud and Attack Model
14.3.2 Deep Neural Network
14.3.3 Performance Metrics
14.3.1.1 Accuracy
14.3.1.2 Precision
14.3.1.3 Recall
14.3.1.4 F1-Score
14.4 Introduction Cloud-Based IDS Module
14.4.1 CNN Model Training
14.4.2 Decision Making and Evaluation
14.4.3 Performance Evaluation
14.4.4 GT-Based Evaluation
14.4.4.1 Game Model Definition
14.4.4.2 Payoff Function
14.5 Discussion and Limitations
14.6 Conclusion
References
Index