1st Edition

Advances in Visual Data Compression and Communication Meeting the Requirements of New Applications

By Feng Wu Copyright 2015
    513 Pages 198 B/W Illustrations
    by Auerbach Publications

    Visual information is one of the richest and most bandwidth-consuming modes of communication. To meet the requirements of emerging applications, powerful data compression and transmission techniques are required to achieve highly efficient communication, even in the presence of growing communication channels that offer increased bandwidth.

    Presenting the results of the author’s years of research on visual data compression and transmission, Advances in Visual Data Compression and Communication: Meeting the Requirements of New Applications provides a theoretical and technical basis for advanced research on visual data compression and communication.

    The book studies the drifting problem in scalable video coding, analyzes the reasons causing the problem, and proposes various solutions to the problem. It explores the author’s Barbell-based lifting coding scheme that has been adopted as common software by MPEG. It also proposes a unified framework for deriving a directional transform from the nondirectional counterpart. The structure of the framework and the statistic distribution of coefficients are similar to those of the nondirectional transforms, which facilitates subsequent entropy coding.

    Exploring the visual correlation that exists in media, the text extends the current coding framework from different aspects, including advanced image synthesis—from description and reconstruction to organizing correlated images as a pseudo sequence. It explains how to apply compressive sensing to solve the data compression problem during transmission and covers novel research on compressive sensor data gathering, random projection codes, and compressive modulation.

    For analog and digital transmission technologies, the book develops the pseudo-analog transmission for media and explores cutting-edge research on distributed pseudo-analog transmission, denoising in pseudo-analog transmission, and supporting MIMO. It concludes by considering emerging developments of information theory for future applications.

    Acronyms

    BASIS FOR COMPRESSION AND COMMUNICATION

    Information Theory
    Introduction
    Source Coding
         Huffman Coding
         Arithmetic Coding
         Rate Distortion Theory
    Channel Coding
         Capacity
         Coding Theorem
         Hamming Codes
    Joint Source and Channel Coding

    Hybrid Video Coding
    Hybrid Coding Framework
    Technical Evolution
         H.261
         MPEG-1
         MPEG-2
         MPEG-4
         H.264/MPEG-4 AVC
         HEVC
         Performance versus Encoding Complexity
    H.264 Standard
         Motion Compensation
         Intra Prediction
         Transform and Quantization
         Entropy Coding
         Deblocking Filtering
         Rate Distortion Optimization
    HEVC Standard
         Motion Compensation
         Intra Prediction
         Transform and Quantization
         Sample Adaptive Offset Filter

    Communication
    Analog Communication
         Analog Modulation
         Multiplexing
    Digital Communication
         Low-Density Parity-Check (LDPC) Codes
         Turbo Codes
         Digital Modulation

    SCALABLE VIDEO CODING

    Progressive Fine Granularity Scalable (PFGS) Coding
    Introduction
    Fine Granularity Scalable Video Coding
    Basic PFGS Framework
         Basic Ideas to Build the PFGS Framework
         The Simplified PFGS Framework
    Improvements to the PFGS Framework
         Potential Coding Inefficiency Due to Two References
         A More Efficient PFGS Framework
    Implementation of the PFGS Encoder and Decoder
    Experimental Results and Analyses
    Simulation of Streaming PFGS Video over Wireless Channels
    Summary

    Motion Threading for 3D Wavelet Coding
    Introduction
    Motion Threading
    Advanced Motion Threading
         Lifting-Based Motion Threading
         Many-to-One Mapping and Non-Referred Pixels
    Multi-Layer Motion-Threading
    Correlated Motion Estimation with R-D Optimization
         Definition of the Mode Types
         R-D Optimized Mode Decision
    Experimental Results
         Coding Performance Comparison  
         Macroblock Mode Distribution
    Summary

    Barbell-Lifting Based 3D Wavelet Coding
    Introduction
    Barbell-Lifting Coding Scheme
         Barbell Lifting
         Layered Motion Coding
         Entropy Coding in Brief
         Base Layer Embedding
    Comparisons with SVC
         Coding Framework
         Temporal Decorrelation
         Spatial Scalability
         Intra Prediction
    Advances in 3D Wavelet Video Coding
         In-Scale MCTF
         Subband Adaptive MCTF
    Experimental Results
         Comparison with Motion Compensated Embedded Zero Block Coding (MC-EZBC) 
         Comparison with Scalable Video Coding (SVC) for Signal-to-Noise Ratio (SNR) Scalability
         Comparison with SVC for Combined Scalability
    Summary

    PART III DIRECTIONAL TRANSFORMS

    DirectionalWavelet Transform
    Introduction
    2D Wavelet Transform via Adaptive Directional Lifting
         ADL Structure
         Subpixel Interpolation
    R-D Optimized Segmentation for ADL
    Experimental Results and Observations
    Summary

    Directional DCT Transform
    Introduction
    Lifting-Based Directional DCT-Like Transform
         Lifting Structure of Discrete Cosine Transform (DCT)
         Directional DCT-Like transform
         Comparison with Rotated DCT
    Image Coding with Proposed Directional Transform
         Direction Transition on Block Boundary
         Direction Selection
    Experimental Results
    Summary

    Directional Filtering Transform
    Introduction
    Adaptive Directional Lifting-Based 2D Wavelet Transform
    Mathematical Analysis
         Coding Gain of ADL
         Numerical Analysis
    Directional Filtering Transform 
         Proposed Intra-Coding Scheme
         Directional Filtering
         Optional Transform
    Experimental Results
    Summary

    VISION-BASED COMPRESSION

    Edge-Based Inpainting
    Introduction
    The Proposed Framework
    Edge Extraction and Exemplar Selection
    Edge-Based Image Inpainting
         Structure
    Experimental Results
    Summary

    Cloud-Based Image Compression
    Introduction
    Related Work
         Visual Content Generation
         Local Feature Compression
          Image Reconstruction
    The Proposed SIFT-Based Image Coding
    Extraction of Image Description
    Compression of Image Descriptors
         Prediction Evaluation
         Compression of SIFT Descriptors
    Image Reconstruction
         Patch Retrieval
         Patch Transformation
         Patch Stitching
    Experimental Results and Analyses
         Compression Ratio
         Visual Quality
         Highly Correlated Image
          Complexity Analyses
         Comparison with SIFT Feature Vector Coding
    Further Discussion
         Typical Applications
         Limitations
         Future Work
    Summary

    Compression for Cloud Photo Storage

    Introduction
    Related Work
         Image Set Compression
         Local Feature Descriptors
    Proposed Scheme
    Feature-Based Prediction Structure
         Graph Building
         Feature-Based Minimum Spanning Tree
         Prediction Structure
    Feature-Based Inter-Image Prediction
         Feature-Based Geometric Deformations
         Feature-Based Photometric Transformation
         Block-Based Motion Compensation
    Experimental Results 
         Efficiency of Multi-Model Prediction
         Efficiency of Photometric Transformation
         Overall Performance
         Complexity
    Our Conjecture on Cloud Storage
    Summary

    COMPRESSIVE COMMUNICATION

    Compressive Data Gathering
    Introduction
    Related Work
         Conventional Compression
         Distributed Source Coding 
         Compressive Sensing
    Compressive Data Gathering
         Data Gathering
         Data Recovery
    Network Capacity of Compressive Data Gathering
         Network Capacity Analysis
         NS-2 Simulation
    Experiments on Real Data Sets
         CTD Data from the Ocean
         Temperature in the Data Center
    Summary

    Compressive Modulation
    Introduction
    Background
         Rate Adaptation
         Mismatched Decoding Problem
    Compressive Modulation
         Coding and Modulation
         Soft Demodulation and Decoding
         Design RP Codes
    Simulation Study
         Rate Adaptation Performance
          Sensitivity to SNR Estimation
    Testbed Evaluation
         Comparison to Oracle
         Comparison to ADM
    Related Work
         Coded Modulation
         Compressive Sensing
    Summary

    Joint Source and Channel Coding

    Introduction
    Related Work and Background
         Joint Source-Channel Coding
         Coded Modulation
         Rate Adaptation
         Compressive Sensing
    Compressive Modulation (CM) for Sparse Binary Sources
         Design Principles
         Weight Selection
         Encoding Matrix Construction
    Belief Propagation Decoding
    Performance Evaluation
         Implementation
         Simulations over an AWGN Channel
         Emulation in Real Channel Environment
    Summary

    PSEUDO-ANALOG tRANSMISSION

    DCast: Distributed Video Multicast
    Introduction
    Related Works
         Distributed Video Coding
         Distributed Video Transmission
         SoftCast
    Proposed DCast
         Coset Coding
         Coset Quantization
         Power Allocation
         Packaging and Transmission
         LMMSE Decoding
    Power-Distortion Optimization
         Relationship between Variables
         MV Transmission Power and Distortion
         MV Distortion and Prediction Noise Variance
          Distortion Formulation
         Solution
    Experiments
         PDO Model Verification
          Unicast Performance
         Evaluation of Each Module
         Robustness Test
         Multicast Performance
         Complexity and Bit-Rate
    Summary

    Denoising in Communication
    Introduction
    Background
         Image Denoising
         Video Compression
    System Design
         System Overview 
         Sender Design
         Receiver Design
    Implementation
         Cactus Implementation 
         GPU Implementation of BM3D
    Evaluation
         Settings
         Micro-Benchmarks
         Comparison against Reference Systems
         Transmitting High-Definition Videos
         Robustness to Packet Loss
    Related Work
    Summary

    MIMO Broadcasting with Receiver Antenna Heterogeneity

    Introduction
    Background and Related Work
         Multi-Antenna Systems
         Layered Source-Channel Schemes
         Compressive Sensing
         SoftCast
    Compressive Image Broadcasting System
         The Encoder and Decoder
         Addressing Heterogeneity
    Power Allocation
         Power Scaling Factors
         Aggregating Coefficients
    Compressive Sampling
    Amplitude Modulation and Transmission
    The CS Decoder
    Simulation Evaluation
         Micro-Benchmarks for Our System
         Performance Comparison with Other Broadcast Systems
    Summary

    FUTURE WORK

    Computational Information Theory
    Introduction
    Cloud Sources
    Source Coding
         Coding of Metadata
         Coding of Cloud Image Sources
         Coding of Cloud Video Sources
         Distributed Coding Using Cloud Sources
    Channel Coding
         Power Allocation and Bandwidth Matching
         Multiple Level Channel Coding
         Channel Denoising
    Joint Source and Channel Coding
    Summary

    Appendix:

    Published Journal and Conference Papers Related to This Book
    Scalable Video Coding
    Directional Transforms
    Vision-Based Compression
    Compressive Communication
    Pseudo-Analog Transmission

    References
    Index

    Biography

    Feng Wu