EE 5359 Project report Low complexity AVS M by implementing data mining algorithm Multimedia Processing EE 5359 Fall 2009 Project Proposal Title Low complexity AVS M codec by implementing datamining algorithm Submitted by Ramolia Pragnesh Guided by Dr K R Rao Dr Dongil Han 1 EE 5359 Project report Low complexity AVS M by implementing data mining algorithm Acknowledgement I would like to thank Dr Rao and Dr Dongil Han for all their guidance and support It was the motivation given by them which led to the successful completion of this project I would also like to thank Dr Kim for helping me in all the ways he could And finally I would like to thank all my friends without whose support it would not have possible for me to complete this project 2 EE 5359 Project report Low complexity AVS M by implementing data mining algorithm Contents 1 Acronyms 4 2 List of figures 5 3 List of tables 6 4 Background and motivation 7 5 Introduction 8 5 1 Introduction to AVS 8 5 2 Introduction to AVS M 10 6 Data structure of AVS M 12 7 Block diagram 16 8 Profile and levels 17 9 Block mode prediction 18 9 1 Intra prediction 18 9 2 Inter prediction 21 10 R D optimization 22 11 Transform quantization and entropy coding 23 12 Machine learning 25 13 Tools used 29 14 13 1 Weka tool 29 13 2 Microsoft Visual Studio 2008 29 Results 30 3 EE 5359 Project report Low complexity AVS M by implementing data mining algorithm 1 Acronyms ARFF AVS M DIP Attributes Relation File Format Audio Video Standard for Mobile Direct Intra Prediction GOP Group Of Picture ICT Integer Cosine Transform IMS Internet Protocol Multimedia Systems IP Internet Protocol MB Macro Block MC Motion Compensation MPEG Moving Picture Experts Group MV Motion Vector PBP Padding before Prediction Weka Waikato Environment for Knowledge Analysis Q Quantization QP Quantization Parameter QCIF Quarter Common Intermediate Format SCI Simplified Chrominance Intra prediction 4 EE 5359 Project report Low complexity AVS M by implementing data mining algorithm TSFT Two Steps Four Taps VLC Variable Length Coding 2 List of figures Figure 1 History of A V coding standard 8 Figure 2 Standard Structure of AVS Video 9 Figure 3 Vertical and horizontal locations of 4 2 0 Y and U V samples in a frame 13 Figure 4 Slice 14 Figure 5 Macro block structure 14 Figure 6 Scanning order in a macro block 15 Figure 7 Data structre of AVS M 15 Figure 8 Block diagram of AVS M encoder 16 Figure 9 Block diagram of AVS M Decoder 17 Figure 10 Intra prediction modes 19 Figure 11 Intra chroma prediction modes in AVS M 19 Figure 12 The position of integer half and quarter pixel samples 21 Figure 13 Inter prediction block sizes 23 Figure 14 Zigzag scanning pattern used for quantized transform coefficients 25 Figure 15 Look how of the tree generated by weka tool 28 Figure 16 45th frame of Akiyo qcif sequence 31 Figure 17 50th frame of foreman cif sequence 31 5 EE 5359 Project report Low complexity AVS M by implementing data mining algorithm Figure 18 50th frame of foreman qcif sequence 31 Figure 19 Screen shot of arff file 32 Figure 20 An example of tree generated by weka tool 33 3 List of tables Table 1 different parts of AVS M 10 Table 2 most probable mode table 20 Table 3 Kth order exponential golomb coding 24 Table 4 input parameters defined for encoding sequence 30 Table 5 AVS M performance 34 6 EE 5359 Project report Low complexity AVS M by implementing data mining algorithm 4 Back ground and motivation The recent deployment of Internet and mobile networks has greatly contributed to the success and adoption of distributed multimedia communications Advancement in today s digital technology has led to research in digital techniques for encoding and decoding process Video trafficking is growing not only with respect to internet but also in mobile communication during recent years Furthermore HDTV broadcasting applications are widely used due to high resolution and good video quality while the image of standard television transmits with a resolution of 720 x 576 pixels the highresolution image has been increased up to 1920 x1080 pixels which implies a considerable increase in data for transmitting Since bandwidth and power requirements are important factors to be considered during transmission Even though it is possible to transmit greater data rates over current communications platforms it is economically not viable to dedicate such an amount of bandwidth to a single video communication service Thus there is a need for video compression which reduces the amount of data to be transmitted and also save some storage space while ensuring high video quality to its customers in compact form AVS M video coding standard is the latest block oriented motion compensation based codec standard developed by the AVS work group of China AVS china can achieve considerably higher coding efficiency than many of the previous standards Unfortunately this comes at a cost in considerably increased complexity at the encoder mainly due to motion estimation and mode decision The high computational complexity of AVS china and real time requirements of video systems represent the main challenge to overcome on the development of efficient encoder solutions The computational power used in AVS M for deciding the prediction mode in inter frame coding is large Say for example assuming FS full search and M block types N reference frames and a search range for each reference frame and block type equal to W we need to examine N x M x 2W 1 2 positions 2 for each block If this computation can be saved it will help reduce much of the load on encoder side making it faster I am proposing a technique to reduce complexity in AVS M using machine learning techniques The idea behind using machine learning is to exploit structural similarities in video in order to make optimal prediction modes through the use 7 EE 5359 Project report Low complexity AVS M by implementing data mining algorithm of fast if else statements instead of the usual cumbersome Sum of Absolute Differences SAD and cost evaluations 5 Introduction 5 1 Introductions to AVS AVS China was developed by the AVS workgroup and is currently owned by China This audio and video standard was initiated by the Chinese government in order to counter the monopoly of the moving picture experts group MPEG standards 27 AVS is a set of integrated standard system which contains systems video audio and media copyright management Different parts of audio video standard AVS and
View Full Document
Unlocking...