The rapid evolution of multimedia communications relying on services and applications that mostly use image and video in different formats has been driving worldwide research and engineering developments towards efficient coding of visual information. Three-dimensional (3D) visual information represented by color texture plus depth is one of such formats, which imposes quite demanding requirements on both hardware and software implementations of standard coding algorithms. Computational complexity, memory, bandwidth, and power consumption are constraining parameters with a great deal of impact on the performance of 3D video coding systems.
This book addresses this challenging field of research and engineering by presenting fast and efficient methods for encoding depth data of 3D visual information within the scope of the 3D extension of the high efficiency video coding (3D-HEVC). The book contains accurate technical descriptions of the main coding tools used for efficient compression of depth maps with particular emphasis on the prediction modes that are responsible for attaining significant compression performance. Jointly with the presentation of the test conditions that must be followed for benchmarking and validation of results, this is crucial and updated information about the standard that provides the necessary background for further analysis and technical developments. Several fast encoding algorithms are described throughout the book, departing from a deep analysis of the main contributing factors for the computational complexity and then proposing efficient solutions to speed up the coding process of depth information at negligible loss of efficiency.
The authors of the book have long research experience and significant contributions for the optimization of standard video coding systems. This highly specialized background also ensures that new methods described in the book are timely and useful for students and researchers interested in further investigating fast coding methods for 3D-HEVC algorithms and engineers developing efficient systems and applications, specially those including 3D multimedia in resource-constrained environments. Overall, this book presents valuable contributions to advance the technical design and implementation of fast and efficient 3D-HEVC standard encoders.

1 Introduction 1

1.1 3D Video With Depth Map Coding 1

1.2 Book Organization and Contributions 4

References 5

2 3D-HEVC Background 7

2.1 3D-HEVC Basic Encoding Structure 7

      2.1.1 3D-HEVC Temporal and Spatial Prediction Structure 7

      2.1.2 Quadtree Structure of 3D-HEVC 9

2.2 Intra-Frame Prediction of Depth Maps 10

      2.2.1 HEVC Intra-Frame Prediction 11

      2.2.2 The Bipartition Modes 12

      2.2.3 Intra-Picture Skip 17

2.3 Inter-Frame and Inter-View Predictions 19

2.4 Common Encoding Steps 20

      2.4.1 Transform-Quantization (TQ) 20

      2.4.2 DC-Only Coding 21

      2.4.3 Entropy Coding 21

2.5 Evaluation Methodology – 3D-HEVC Common Test Conditions (CTC) 21

References 24

3 State-of-the-Art Overview 27

3.1 State-of-the-Art Concerning the Reduction of the Computational Effort Applied to the Intra-Frame Prediction 27

      3.1.1 State-of-the-Art of the Mode-Level Decision Algorithms 28

      3.1.2 State-of-the-Art in Block-Level Decision Algorithms 29

      3.1.3 State-of-the-Art in Quadtree-Level Decision Algorithms 30

      3.1.4 Summary of the State-of-the-Art in Intra-Frame Prediction 31

3.2 State-of-the-Art in Reducing the Computational Effort on the Inter-Frame/View Prediction 32

References 33

4 Encoding Time Reduction for 3D-HEVC Intra-Frame Prediction of Depth Maps 35

4.1 Performance Analysis 35

      4.1.1 Encoding Time Distribution 35

      4.1.2 Block Sizes and Encoding Mode Distribution 40

4.2 Algorithms Designed for the Intra-Frame Prediction of the 3D-HEVC Depth Maps Coding 43

      4.2.1 DMM-1 Fast Pattern Selector (DFPS) 43

      4.2.2 Pattern-Based Gradient Mode One Filter (P&GMOF) 49

      4.2.3 Enhanced Depth Rough Mode Decision (ED-RMD) 52

      4.2.4 Applying Machine Learning for Quadtree Limitation 54

4.3 Experimental Results 57

References 61

5 Novel Encoding Time Reduction Algorithms for the 3D-HEVC Inter-Frame Prediction 63

5.1 Edge-Aware Depth Motion Estimation (E-ADME) 63

      5.1.1 Designed E-ADME Scheme 64

5.2 Early Termination Block-Level Decision Algorithm 65

      5.2.1 Design of the Early Termination Block-Level Decision 67

5.3 Decision Trees for P- and B-Frames 69

5.4 Results and Comparisons 72

References 76

6 Conclusions and Open Research Possibilities 77

6.1 Future Works and Open Research Possibilities 78

Index 79

中科院文献情报中心