Video Coding with Superimposed Motion-Compensated Signals: Applications to H. 264 and Beyond

Flierl M., Girod B.1402077653

Video Coding with Superimposed Motion-Compensated Signals: Applications to H.264 and Beyond captures recent advances in motion compensation for efficient video compression. This book investigates linearly combined motion compensated signals, and generalizes the well known superposition for bidirectional prediction in B-pictures. The number of superimposed signals and the reference picture selection are important aspects of discussion.The application oriented part of the book employs the concept to ITU-T Recommendation H.263, and continues with the improvements by superimposed motion-compensated signals for the EMERGING STANDARDS ITU-T Recommendation H.264 & ISO/IEC MPEG-4 part 10. In addition, Video Coding with Superimposed Motion-Compensated Signals: Applications to H.264 and Beyond discusses a new approach for wavelet-based video coding. This technology is currently being investigated by MPEG, to develop a new video compression standard for the mid-term future.The theoretical part of Video Coding with Superimposed Motion-Compensated Signals: Applications to H.264 and Beyond provides a deeper understanding of the underlying principles of superimposed motion-compensated signals. This book incorporates more than 200 references, summarizes relevant prior work, and develops a mathematical characterization of superimposed motion-compensated signals. The derived information-theoretic performance bounds permit a valuable comparison of the investigated compression schemes.Video Coding with Superimposed Motion-Compensated Signals: Applications to H.264 and Beyond is designed for aprofessional audience composed of researchers and practitioners in industry. This book is also suitable as a secondary text for graduate-level students in computer science and engineering.

---

Table of contents :
VIDEO CODING WITH SUPERIMPOSED MOTION-COMPENSATED SIGNALS……Page 1
Contents……Page 6
Preface……Page 10
Acknowledgments……Page 12
1. INTRODUCTION……Page 14
2.1 Coding of Video Signals……Page 18
2.2 Motion Compensation……Page 19
2.2.1 Bidirectional Motion Compensation……Page 20
2.2.2 Overlapped Block Motion Compensation……Page 22
2.2.3 Variable Block Size Motion Compensation……Page 23
2.2.4 Multiframe Motion Compensation……Page 24
2.2.5 Superimposed Motion Compensation……Page 25
2.3 Motion Estimation……Page 28
2.3.1 Rate-Constrained Motion Estimation……Page 29
2.3.2 Rate-Constrained Estimation of Superimposed Motion……Page 31
2.3.3 Quantizer Selection at the Residual Encoder……Page 34
2.3.4 Efficient Motion Estimation……Page 35
2.4 Theory of Motion-Compensated Prediction……Page 36
2.4.1 Frame Signal Model……Page 37
2.4.2 Signal Model for Motion-Compensated Prediction……Page 39
2.4.3 Signal Model for Multihypothesis Prediction……Page 40
2.4.4 Performance Measures……Page 42
2.4.5 Conclusions……Page 43
2.5.1 Motion Compensation and Subband Coding……Page 44
2.5.2 Motion-Compensated Lifted Wavelets……Page 46
3.1 Introduction……Page 48
3.2.1 Superimposed Prediction and Correlated Displacement Error……Page 49
3.2.2 Complementary Hypotheses……Page 52
3.2.3 Gradient of the Prediction Error Variance……Page 57
3.3 Hypotheses with Additive Noise……Page 59
3.3.1 Averaging Filter……Page 60
3.3.2 Wiener Filter……Page 62
3.4 Forward-Adaptive Hypothesis Switching……Page 69
3.4.1 Signal Model for Hypothesis Switching……Page 70
3.4.2 Minimizing the Radial Displacement Error……Page 71
3.4.3 Equivalent Predictor……Page 74
3.4.4 Motion Compensation with Complementary Hypotheses and Forward-Adaptive Hypothesis Switching……Page 76
3.5 Pictures with Varying Number of Hypotheses……Page 78
3.6 Conclusions……Page 80
4.1 Introduction……Page 82
4.2.1 Syntax Extensions……Page 83
4.2.2 Coder Control……Page 84
4.3 Experimental Results……Page 85
4.3.1 Multiple Hypotheses for Constant Block Size……Page 86
4.3.2 Multiple Hypotheses for Variable Block Size……Page 87
4.3.3 Multiple Hypotheses and Multiple Reference Pictures……Page 90
4.4 Conclusions……Page 96
5.1 Introduction……Page 98
5.2.1 Overview……Page 100
5.2.2 Direct Mode……Page 101
5.2.4 Rate-Distortion Performance of Individual Modes……Page 102
5.3.1 Bidirectional vs. Superposition Mode……Page 105
5.3.2 Two Combined Forward Prediction Signals……Page 108
5.3.3 Entropy Coding……Page 110
5.4.1 Rate-Constrained Mode Decision……Page 113
5.4.2 Rate-Constrained Motion Estimation……Page 114
5.4.3 Rate-Constrained Superimposed Motion Estimation……Page 115
5.4.4 Improving Overall Rate-Distortion Performance……Page 117
5.5 Conclusions……Page 120
6.1 Introduction……Page 122
6.2.1 Motion-Compensated Lifted Haar Wavelet……Page 123
6.2.2 Motion-Compensated Lifted 5/3 Wavelet……Page 124
6.2.3 Experimental Results……Page 125
6.3.1 Motion-Compensated Lifted Haar Wavelet……Page 131
6.3.2 Motion-Compensated Lifted 5/3 Wavelet……Page 133
6.3.3 Signal Model……Page 134
6.3.4 Transform Coding Gain……Page 137
6.4 Conclusions……Page 140
7. SUMMARY……Page 142
Singularities of the Displacement Error Covariance Matrix A.1……Page 146
Inverse of the Power Spectral Density Matrix A.3……Page 147
Power Spectral Density of a Frame A.4……Page 149
Glossary……Page 150
References……Page 154
Index……Page 172