This preview shows page 1-2-3-4-5-6-7-8-9-63-64-65-66-67-68-69-70-71-127-128-129-130-131-132-133-134-135 out of 135 pages.
IMPLEMENTATION AND PERFORMANCE ANALYSIS OF 2 D ORDER 16 INTEGER TRANSFORMS IN H 264 AVC AND AVS VIDEO FOR HIGH DEFENITION VIDEO CODING by MADHU PERINGASSERY KRISHNAN Presented to the Faculty of the Graduate School of The University of Texas at Arlington in Partial Fulfillment of the Requirements for the Degree of MASTER OF SCIENCE IN ELECTRICAL ENGINEERING THE UNIVERSITY OF TEXAS AT ARLINGTON DECEMBER 2010 Copyright by Madhu Peringassery Krishnan 2010 All Rights Reserved ACKNOWLEDGEMENTS I would like to thank my advisor Dr K R Rao for his guidance support and encouragement at every stage of development in this research work I would like to thank Dr Manry and Dr Davis for serving on my committee I would like to express my gratitude to Dr P Topiwala for his inputs towards the completion of this research Last but not least I would like to thank my family and friends who have helped me throughout my life November 24 2010 iv ABSTRACT IMPLEMENTATION AND PERFORMANCE ANALYSIS OF 2 D ORDER 16 INTEGER TRANSFORMS IN H 264 AVC AND AVS VIDEO FOR HIGH DEFENITION VIDEO CODING Madhu Peringassery Krishnan M S The University of Texas at Arlington 2010 Supervising Professor K R Rao H 264 AVC and AVS video are two video coding standards that have a wide range of applications ranging from high end professional camera and editing systems to low end mobile applications They strive to achieve maximum compression efficiency without compromising the quality of video To this end many coding tools are defined in them Transform coding is one among them Transform coding represents the signal image that is currently in time spatial domain in another domain transform domain where most of the energy in signal image is concentrated in a fewer number of coefficients Thus the insignificant coefficients can be discarded after transform coding to achieve compression In images videos the DCT II which represents a signal image as the weighted sum of cosine functions with different frequencies is primarily used for transform coding v H 264 AVC and AVS video utilize integer approximations of the DCT II known as integer cosine transform to reduce computational complexity by performing only fixed point arithmetic operations and eliminates the mismatches between the forward and inverse transforms The order size of the integer cosine transforms used is small 4 x 4 and 8 x 8 They achieve the best coding efficiency for standard definition and low resolution videos But better coding efficiency can be achieved for high definition videos by using higher order 16 x 16 and 32 x 32 integer cosine transforms As high definition videos are becoming more and more popular it is imperative that sooner or later they will be integrated into the standards For this purpose many higher order 16 x 16 and 32 x 32 integer cosine transforms have been proposed But a comparative study on the performance of these higher order integer cosine transforms in H 264 AVC and AVS video has not been done yet The purpose of this research is to analyze some higher order 16 x 16 integer cosine transforms implement them in H 264 AVC and AVS video and carry out a comparative study of their performances vi TABLE OF CONTENTS ACKNOWLEDGEMENTS iv ABSTRACT v LIST OF ILLUSTRATIONS x LIST OF TABLES xiv Chapter Page 1 INTRODUCTION 20 1 1 Discrete cosine transform and video compression 20 1 2 Integer cosine transforms 21 1 3 HD video coding and integer cosine transforms 23 1 4 Outline 26 2 H 264 AVC 27 2 1 Introduction 27 2 2 H 264 AVC encoder 30 2 2 1 Intra prediction 31 2 2 2 Inter prediction 34 2 2 3 Transform coding 37 2 2 4 Entropy coding 40 2 2 5 Deblocking filter 40 2 2 6 Error resilience 41 2 3 H 264 AVC decoder 41 vii 3 AVS VIDEO 43 3 1 Introduction 43 3 2 AVS video encoder 43 3 2 1 Intra prediction 44 3 2 2 Inter prediction 47 3 2 3 Transform coding 50 3 2 4 Quantization and scanning 51 3 2 5 Entropy coding 53 3 2 6 In loop Deblocking filter 54 3 2 7 Error resilience 55 3 3 AVS video encoder 55 4 HIGHER ORDER 2 D ICTs FOR HD VIDEO CODING 57 4 1 Introduction 57 4 2 Integer cosine transforms 58 4 3 Simple 2 D order 16 ICT 59 4 4 Modified 2 D order 16 ICT 66 4 5 2 D order 16 binDCT based on Loeffler s factorization 65 5 PERFORMANCE ANALYSIS AND CONCLUSION 78 5 1 Introduction 78 5 2 Implementation in H 264 AVC and performance analysis 80 5 2 1 Performance of simple 2 D order 16 ICT SICT 81 5 2 2 Performance of Modified 2 D order 16 ICT MICT 87 5 2 3 Performance of 2 D order 16 binDCT L 93 5 3 Implementation in AVS video and performance analysis 99 viii 5 3 1 Performance of simple 2 D order 16 ICT SICT 100 5 3 2 Performance of Modified 2 D order 16 ICT MICT 107 5 3 3 Performance of 2 D order 16 binDCT L 112 5 4 Conclusions and future work 118 APPENDIX A THE 16 x 16 MATRICES FOR QUANTIZATION AND DEQUANTIZATION 120 B SELECTED FRAMES FROM VIDEO SEQUENCES 123 REFERENCES 130 BIOGRAPHICAL INFORMATION 135 ix LIST OF ILLUSTRATIONS Figure Page 1 1 Separable property of DCT II 20 1 2 Typical block diagram of a video encoder 22 1 3 Typical block diagram of a video decoder 22 1 4 ICTs in H 264 AVC a 4 x 4 transform matrix b 8 x 8 transform matrix 24 1 5 8 x 8 ICT matrix in AVS video 25 2 1 Profiles in H 264 AVC 29 2 2 Single frame motion compensated prediction 30 2 3 H 264 AVC encoder block representation 31 2 4 Grouping of MBs into slices 32 2 5 Nine prediction directions for 4 x 4 blocks 33 2 6 Nine intra prediction modes for 4 x 4 and 8 x 8 blocks 34 2 7 Partitioning of MBs for motion compensated prediction 34 2 8 Multi frame motion compensated prediction 35 2 9 a Predicting half sample positions b Predicting quarter sample positions from the half samples 36 2 10 The 4 x 4 integer transform 37 2 11 a The sixteen 4 4 blocks with DC coefficients drawn as smaller blocks on upper left corner b The 4 4 Hadamard and the 2 2 Haar Hadamard transforms applied on the DC coefficients 38 2 12 The 8 x 8 integer transform matrix 39 2 13 Zigzag scan 39 x 2 14 Luma and chroma boundaries to be filtered by in loop deblocking filter 41 2 15 H 264 AVC decoder block diagram 42 3 1 AVS video encoder 44 3 2 Five prediction modes for 8 8 luma blocks directions for 8 8 intra prediction upper left mode 3 upper right mode 0 middle left mode 2 middle right mode 1 bottom left mode 4 bottom right 45 3 3 a Directional modes for …
View Full Document
Unlocking...