Smart Algorithms for Multimedia and Imaging(2021)[Rychagov et al][9783030667412]

Preface
Acknowledgment
Contents
About the Editors
Chapter 1: Super-Resolution: 1. Multi-Frame-Based Approach
	1.1 Super-Resolution Problem
		1.1.1 Introduction
		1.1.2 Super-Resolution and Interpolation, Image Formation Model
		1.1.3 Optimization Criteria
	1.2 Fast Approaches to Super-Resolution
		1.2.1 Problem Formulation Allowing Fast Implementation
		1.2.2 Block Diagonalization in Super-Resolution Problems
		1.2.3 Filter-Bank Implementation
		1.2.4 Symmetry Properties
		1.2.5 Discussion of Results
	1.3 Practical Super-Resolution
		1.3.1 System Architecture
		1.3.2 Structure Tensor in Bayer Domain and Salience Map
		1.3.3 Motion Estimation in Bayer Domain
	References
Chapter 2: Super-Resolution: 2. Machine Learning-Based Approach
	2.1 Introduction
	2.2 Training Datasets for Super-Resolution
	2.3 Loss Functions and Image Quality Metrics
	2.4 Super-Resolution Implementation on a Mobile Device
	2.5 Super-Resolution Competitions
	2.6 Prominent Deep Learning Models for Super-Resolution
	2.7 Notable Applications and Future Challenges
	References
Chapter 3: Depth Estimation and Control
	3.1 Introduction
	3.2 Stereo Content Reproduction Systems
		3.2.1 Passive System with Polarised Glasses
		3.2.2 Active Shutter Glass System
		3.2.3 Colour-Based Views Separation
	3.3 Eye Fatigue
	3.4 Depth Control for Stereo Content Reproduction
	3.5 Fast Recursive Algorithm for Depth Estimation From Stereo
	3.6 Depth Post-Processing
	3.7 Stereo Synthesis
		3.7.1 Disocclusion
		3.7.2 Temporal Consistency
		3.7.3 Symmetric Versus Asymmetric View Generation
		3.7.4 Toed-in Camera Configuration
	References
Chapter 4: Semi-Automatic 2D to 3D Video Conversion
	4.1 2D to 3D Semi-automatic Video Conversion
	4.2 Video Analysis and Key Frame Detection
	4.3 Depth Propagation from Key Frames
		4.3.1 Introduction
		4.3.2 Related Work
		4.3.3 Depth Propagation Algorithm
		4.3.4 Results
	4.4 Motion Vector Estimation
		4.4.1 Introduction
		4.4.2 Variational Optical Flow
		4.4.3 Total Variation Optical Flow Using Two Colour Channels
		4.4.4 Total Variation Optical Flow with a Non-local Smoothness Term
		4.4.5 Special Weighting Method for Fast Non-local Optical Flow Computation
		4.4.6 Solver
		4.4.7 Evolution of Data Term Importance
		4.4.8 Clustering of Motion
		4.4.9 Results
	4.5 Background Inpainting
		4.5.1 Introduction
		4.5.2 Related Work
		4.5.3 Background Inpainting
		4.5.4 Results
	4.6 View Rendering
	4.7 Stereo Content Quality
	References
Chapter 5: Visually Lossless Colour Compression Technology
	5.1 Why Embedded Compression Is Required
		5.1.1 Introduction
		5.1.2 Prior-Art Algorithms and Texture Compression Technologies
	5.2 Requirements and Architecture of VLCCT
	5.3 Methods for Encoding
		5.3.1 Overview of 2 x 4 Methods
		5.3.2 Overview of 2 x 2 Methods
	5.4 How to Pick the Right Method
	5.5 Bitstream Syntax and Signalling
	5.6 Complexity Analysis
	5.7 Quality Analysis
	5.8 Conclusions
	References
Chapter 6: Automatic Video Editing
	6.1 Related Work
		6.1.1 Introduction
		6.1.2 Film Theory Studies
		6.1.3 Video Summarising
		6.1.4 Automated Editing of Video Footage from Multiple Cameras to Create a Coherent Narrative of an Event
		6.1.5 Automatic Editing of Single-Camera, Multiple-Angle and Multiple-Take Footage to Create a Coherent Final Scene
		6.1.6 Learning Editing Styles from Existing Films
		6.1.7 Generating Video from Video Clips Based on Moments of Interest
		6.1.8 Nonparametric Probabilistic Approach for Media Analysis
	6.2 Imitation Learning
		6.2.1 Solution Overview
		6.2.2 Features Extraction
		6.2.3 Automatic Editing Model Training
		6.2.4 Qualitative Evaluation
		6.2.5 Quantitative Evaluation
	6.3 Dynamic Programming
		6.3.1 Problem Statement
		6.3.2 Montage Transitions and Frame Image Quality Scoring
		6.3.3 Cost Function
		6.3.4 Implementation Details
		6.3.5 Automatic Editing Results Evaluation
	6.4 Conclusion
	References
Chapter 7: Real-Time Detection of Sports Broadcasts Using Video Content Analysis
	7.1 Genre Detection as a Tool for Video Adjustment
		7.1.1 Introduction
		7.1.2 Video Classification in Brief
		7.1.3 Requirements for a Real-Time Video Classification Algorithm
	7.2 Description of the Algorithm
	7.3 Results
		7.3.1 Basic Classification
		7.3.2 Processing of Corner Cases
	7.4 Video Genre Classification: Recent Approaches and CNN-Based Algorithms
	References
Chapter 8: Natural Effect Generation and Reproduction
	8.1 Introduction
	8.2 Previous Works
	8.3 Animation Effects from a Photo
		8.3.1 General Workflow
		8.3.2 Flashing Light
		8.3.3 Soap Bubbles
		8.3.4 Sunlight Spot
	8.4 Adaptation to Audio
	8.5 Results and Discussion
	8.6 Future Work
	References
Chapter 9: Image Classification as a Service
	9.1 Introduction
		9.1.1 Types of Image Classification Systems
		9.1.2 Constraints and Assumptions
	9.2 Ethical Considerations
	9.3 Metrics and Evaluation
		9.3.1 End-to-End Metrics
		9.3.2 Classification Metrics
		9.3.3 Training and Inference Performance Metrics
		9.3.4 Uncertainty and Robustness Metrics
	9.4 Data
		9.4.1 Data Acquisition
		9.4.2 Human Annotation or Verification
		9.4.3 Data Engineering
		9.4.4 Data Management
		9.4.5 Data Augmentation
	9.5 Model Training
		9.5.1 Model Architecture
		9.5.2 Classification Model Training Approaches
	9.6 Deployment
	References
Chapter 10: Mobile User Profiling
	10.1 Introduction
	10.2 Demographic Prediction Based on Mobile User Data
		10.2.1 Pre-processing
		10.2.2 Probabilistic Latent Semantic Analysis
		10.2.3 Additive Regularization of Topic Models
		10.2.4 Document Aggregation
	10.3 Behaviour-Based Authentication on Mobile Devices
	10.4 Dataset Collection System: Architecture and Implementation Features
	References
Chapter 11: Automatic View Planning in Magnetic Resonance Imaging
	11.1 View Planning in MRI
		11.1.1 Introduction
		11.1.2 Related Works on AVP
	11.2 Automatic View Planning Framework
		11.2.1 Midsagittal Plane Estimation in Brain Images
		11.2.2 Anatomical Landmark Detection
		11.2.3 Training Landmark Detector
		11.2.4 Spine Vertebrae Alignment
		11.2.5 Landmark Candidate Position Refinement Using Statistical Model
	11.3 Results
	11.4 Conclusion
	References
Chapter 12: Dictionary-Based Compressed Sensing MRI
	12.1 Toward Perfect MRI
		12.1.1 Introduction
	12.2 Compressed Sensing MRI
		12.2.1 Observation Model
		12.2.2 Sparse Object Representation
	12.3 Dictionary-Based CS MRI
		12.3.1 Alternating Optimization
		12.3.2 Dictionary Learning
	12.4 Proposed Efficient MRI Algorithm
		12.4.1 Split Bregman Initialization
		12.4.2 Precomputed Dictionary
		12.4.3 Multi-Band Decomposition
		12.4.4 Non-uniform-Sized Patches
		12.4.5 DESIRE-MRI
		12.4.6 GPU Implementation
	12.5 Experimental Results for Multi-Coil MRI Reconstruction
		12.5.1 Efficiency of DESIRE-MRI Innovations
		12.5.2 DESIRE-MRI Reconstruction Accuracy
		12.5.3 Computational Performance
	12.6 Conclusion
	References
Chapter 13: Depth Camera Based on Colour-Coded Aperture
	13.1 Introduction
	13.2 Existing Approaches and Recent Trends
	13.3 Numerical Simulation and Its Applications
	13.4 Light-Efficient Colour-Coded Aperture Designs
	13.5 Depth Map Estimation for Thin Lens Cameras
	13.6 Depth Sensor Calibration for Complex Objectives
	13.7 Evaluation Results
	13.8 Prototypes and Implementation
	References
Chapter 14: An Animated Graphical Abstract for an Image
	14.1 Introduction
	14.2 Related Work
	14.3 An Animated Graphical Abstract
		14.3.1 General Idea
		14.3.2 Attention Zone Detection for Photos
		14.3.3 Attention Zone Detection for Images of Documents
		14.3.4 Attention Zones for a Slice of a Tomographic Image
		14.3.5 Generation of Animation
	14.4 Results and Discussion
	References
Chapter 15: Real-Time Video Frame-Rate Conversion
	15.1 Introduction
	15.2 Frame-Rate Conversion Algorithm Structure
	15.3 ME Stage: Real-Time 3DRS Algorithm
		15.3.1 Baseline 3DRS-Based Algorithm
		15.3.2 Wave-Front Scanning Order
		15.3.3 Slanted Wave-Front Scanning Order
		15.3.4 Double-Block Processing
		15.3.5 Evaluation of 3DRS Algorithm Modifications
	15.4 OP Stage - Detection of Occlusions and Correction of Motion
	15.5 MCI Stage: Motion-Compensated Frame Interpolation
	15.6 Problematic Situations and Fallback
	15.7 FRC Pipeline: Putting Stages Together
	15.8 Results
	References
Chapter 16: Approaches and Methods to Iris Recognition for Mobile
	16.1 Introduction
	16.2 Person Recognition by Iris
	16.3 Iris Recognition for Mobile Devices
	16.4 Iris Feature Extraction and Matching
	16.5 Limitations of Iris Recognition and Approaches for Shifting Limitations
	References
Index