基於OpenMAX(TM) DL之H.264解碼器在PACDSP數位訊號處理器上的實作與最佳化

簡易檢索 / 詳目顯示

回結果列表

研究生：	林立浩 Li-HaoLin
論文名稱：	基於OpenMAX(TM) DL之H.264解碼器在PACDSP數位訊號處理器上的實作與最佳化 Implementation and Optimization H.264 Decoder based on OpenMAX(TM) DL for PACDSP
指導教授：	石維寬 Wei-Kuan Shih
口試委員:
學位類別：	碩士 Master
系所名稱：	電機資訊學院 - 資訊工程學系 Computer Science
論文出版年：	2007
畢業學年度：	95
語文別：	中文
論文頁數：	90
中文關鍵詞：	林立浩、數位訊號處理器、石維寬、多媒體編碼、多媒體解碼、多媒體
外文關鍵詞：	H.264, OpenMAX, PACDSP, video, LiHao, OpenMAX DL
相關次數：	點閱：2 下載：0
分享至:	分享至facebook 分享至twitter

查詢本校圖書館目錄查詢臺灣博碩士論文知識加值系統勘誤回報

隨著多媒體時代的來臨，以及IC產業的進步,新一代的PMP (Protable Media Player)功能越來越強大，在多媒體功能的要求上是不容置疑的，多媒體種類眾多，包括音樂/音效、語音、照片/圖片(靜態影像)、視訊(動態影像)、2D/3D動畫等。現今的的行動多媒體裝置大多具備幾項能力，一、支援H.264、MPEG-4、MP3、VC1、AAC這幾個目前較主流的影像和音訊編解碼標準，二、支援無線通訊標準，比如在手機平台上加入GSM、GPRS、3G或WiMAX等通訊標準。
上面這些技術需要多種處理能力及複雜的電路設計，如果個別來達到這些技術設計的話，勢必無法達到體積小、攜帶方便的目標。因此整合各種多媒體解碼能力的SOC被廣泛應用在行動多媒體裝置。也因為這樣所以一個行動多媒體裝
置需要程度上硬體及軟體的配合，在效能上以及省電要求上是很重要的兩個考慮議題。目前最受歡迎的PMP架構為多核心處理器，一顆MPU(Main Processor Unit)搭配一顆或多顆DSP(Digital Signal Processor) 。
OpenMAXTM為一套由Khronos組織發佈的統一標準，其中有兩個重要介面，分別是DL(Development Layer)層、IL(Integration Layer)層，主要的優點是透OpenMAXTM訂出的這兩個統一介面能夠讓使用者在處理多媒體編解碼上達到跨平台及加速底層硬體效能。
本篇論文目的主要是能在Dual Core平台上加速多媒體處理效能，實驗平台為工研院的PAC平台，主要會在PACISS模擬器下作測試，多媒體編解碼採用H.264解碼器。而H.264解碼器的實作上則遵循OpenMAXTM DL的標準，再自行設計出一套函式庫，此函式庫是用PACDSP Assembly程式語言實作演算法且經過最佳化，讓上層使用者能夠藉由呼叫設計出來的函式庫，更方便的與底層的PACDSP平台溝通，並且讓硬體效能發揮的更好。

With the arrival of the multimedia, and the improvement of IC industry, the functions of the new generation PMP (portable Media Player) is becoming stronger and stronger. It is no doubt that the powerful functions of multimedia are expected. There are many kinds of multimedia, including music/sound effect, audio, picture (static image), video (dynamic image), 2D/3D animation, etc. The most mobile multimedia devices have several abilities. First, they support H.264, MPEG-4, MP3, VC1, and AAC which are the main stream of image and audio codec. Second, they support the standard of wireless communication, such as GSM, GPRS, 3G, or WiMAX on the cell-phone platform.

These technologies mentioned above need many kinds of process ability and more complicated design of circuit. If each of these technologies is designed individually, it is definitely impossible to reach the goals of small size and the portability. Thus SOC with integration ability is applied on the mobile media device extensively. However, because these devices need the cooperation on hardware and software, two important issues about efficiency and energy saving are worth of being considered. The most popular PMP architecture is multi-core processor that has one MPU (Main Processor Unit) plus one or more DSP (Digital Signal Processor).

OpenMAXTM is a unified standard released by Khronos group. There are two important interfaces. One is DL (Development Layer) and the other one is IL (Integration Layer). The major advantage is that these two interfaces can cross platforms and accelerate the performance of hardwares when the users and programmers process multimedia codec.

The main motivation of this thesis is to accelerate the performance of processing multimedia. The platform of the experiment was called PACDSP platform which was developed by ITRI. The testing was under the environment of PACISS simulator. The multimedia codec adopted H.264 decoder, which followed OpenMAXTM DL standard. Then the author designed one library by PACDSP assembly code and implemented the algorithm as well as optimization. Therefore, the upper users can call this designed library to communicate with the PACDSP hardware more conveniently.

中文摘要    I
Abstract    II
致謝    III
目錄    IV
圖目錄    VII
表格目錄    X
  Introduction    1
1    Background    1
2    Motivation    2
  PACDSP and OpenMAXTM    4
1    PACDSP Overview    4
1.1    Introudction    4
1.1.1    Target Markets    4
1.1.2    Architecture Features    4
1.2    PACDSP Core Architecture    5
1.3    PAC DSP Kernel Architecture    5
1.3.1    Pipeline Stages    6
1.3.2    Program Sequence Control Unit    7
1.3.3    Scalar Unit    7
1.3.4    VLIW Data Path    7
1.3.4.1    Arithmetic Unit    8
1.3.4.2    Load/Store Unit    9
1.3.4.3    VLIW Datapath Register Organization    10
1.4    PACDSP Assembly Programming    10
1.4.1    PACDSP Instruction Set Architecture    10
1.4.2    Syntax of PACDSP Assembly Language    11
1.4.3    Instruction Package    12
1.4.3.1    Instruction Slots    12
1.4.3.2    Instruction Broadcast    12
1.4.4    Coding Constraint    12
2    OpenMAXTM Overview    14
2.1    About the Khronos Group    14
2.2    OpenMAXTM    15
2.3    The OpenMAXTM Integration Layer    16
2.4    The OpenMAXTM Development Layer    17
2.4.1    The Interface    17
2.4.2    Definitions    17
2.4.3    API Conventions    18
2.4.3.1    Function Naming    18
2.4.3.2    Function Arguments    21
2.4.3.3    Data Types    21
  H.264 Codec Followed OpenMAXTM DL    22
1    Introduction    22
2    Video Formats    22
3    H.264/MPEG4 Part 10    23
3.1    Hierarchy of a Video Sequence    24
3.2    The H.264 Codec    25
3.3    H.264 Structure    26
3.3.1    Profiles and Levels    26
3.3.2    Reference Pictures    27
3.3.3    Slices    28
3.3.4    Macroblock Prediction    29
3.4    Decoding Process    29
3.4.1    Intra Prediction    29
3.4.2    Inter Prediction    30
3.4.3    Deblocking Filter    32
3.4.4    Transformation and Quantization    33
3.4.5    Reorder    34
3.4.6    Entropy Coding    35
4    Spilt of H.264 Decoder Followed OpenMAXTM DL    35
  Implementation and Optimization    37
1    Introduction    37
2    Memory Allocation of Reference Data    37
3    Intra Prediction    38
3.1    Intra Luma 4x4    38
3.1.1    Mode 0 (block_vertical)    40
3.1.2    Mode 1 (block_horizontal)    40
3.1.3    Mode 2 (DC)    41
3.1.4    Mode 3 (block_diag_down_left)    42
3.1.5    Mode 4，5，6，7，8    43
3.2    Intra Luma 16x16    44
3.2.1    Mode 0 (block_vertical)    47
3.2.2    Mode 1 (block_horizontal)    47
3.2.3    Mode 2 (block_dc)    48
3.2.4    Mode 3 (block_plane)    49
3.3    Intra Chroma 8x8    51
3.3.1    Mode 0 (block_dc)    53
3.3.2    Mode 1 (block_horizontal)    53
3.3.3    Mode 2 (block_vertical)    54
3.3.4    Mode 3 (block_plane)    55
4    Deblocking Filter    55
4.1    Deblocking Filter Luma Vertical Edge    56
4.2    Deblocking Filter Luma Horizontal Edge    58
4.3    Deblocking Filter Chroma Vertical Edge    60
4.4    Deblocking Filter Chroma Horizontal Edge    61
5    Inverse Transformation and Inverse Quantization    62
5.1    Luma IDCT/Q    62
5.2    Chroma IDCT/Q    68
5.3    IACQ/T    70
6    Inter Prediction    71
6.1    Inter Luma    71
6.2    Inter Chroma    75
  Experiment and Statistic Analysis    77
1    Environment of Experiment    77
2    Comparison between Assembly and Pure C with PAC Compiler    77
3    Cycle about Each Functional Block    81
4    Relation between Register File Utilization and Optimization    82
  Conclusion    84
1    Conclusion    84
2    Suggestion    85
  Future Work    86
Reference    87
Appendix A    89
A.1    Boundary Strength (BS = 4)    89
A.2    Boundary Strength (BS = 1, 2, 3)    89
A.3    Quantization Step    89

                                

[1] STC/E100, “PACDSP v3.0 PACDSP Software Programming Guide” version 0.9.1, December 2006.

[2] Thomas Wiegand, Gary J. Sullivan, Gisle Bjøntegaard, and Ajay Luthra, Senior Member, IEEE,“Overview of the H.264/AVC Video Coding Standard”, IEEE Transactions on Circuits and Systems for Video Technology, Vol 13, No 7, JULY 2003.

[3] E.G. Richardson, “H.264 and MPEG-4 Video Compresion ─ Video Coding for Next-generation Multimedia”UK ISBN0-471-84837-5, John Wiley & sons Press, 2003.

[4] JVT of O/IEC JTC1/SC29/WG11/N6359, “Text of ISO/IEC 14496-10:2004
Advanced Video Coding (second edition)“Munich(DE), March 2004.

[5] Xiaosong Zhou, Eric Q. Li, and Yen-Kuang Chen, ”Implementation of H.264
Decoder on General-Purpose Processors with Media Instructions“, Proceedings of SPIE -- Volume 5022, Image and Video Communications and Processing, May 2003.

[6] The Khronos Group Inc.“OpenMAX Development Layer API Specification”, version 1.0.1, June 30th, 2006.

[7] The Khronos Group In.“OpenMAXTM Integration Layer application programming Interface Specification“, Version 1.1, February 21, 2007.

[8] JianLiang Luo,“Implementation and Optimaization of H.264 baseline profile
decoder on PACDSP dual core platform”NTHU.CS, July, 2006.

[9] Chuan-Hua Chang, I-Tao Liao, “PAC DSP: A Digital Signal Processor for
Next-generation Mobile Platform”, Soc Technical Journal 002, May 2005.

[10] I-Tao Liao, Chan-Hao Chang, Tse-Hao Lee, Chuan-Cheng Peng,“PAC DSP”，
SoC Technial Journal 004，April 2006.

[11]“OpenMAX Sample Code Release Note, OX000-BU-00000-r0p0-00bet0”, ARM, 2006.

[12] OpenMAX. URL: http://www.khronos.org/openmax/

[13] ARM OpenMAX DL Libraries: URL:
http://www.arm.com/products/esd/openmax_armlibraries.html

[14] Gary J. Sullivan, Pankaj Topiwala, Ajay Luthra, “The H.264/AVC Advanced Video Coding Standard: Overview and Introduction to the Fidelity Range Extensions”, SPIE on Applications of Digital Image Processing XXVII Special Session on Advances in the NEW Emerging Standard: H.264/AVC, August, 2004

全文公開日期本全文未授權公開 (校內網路)
全文公開日期本全文未授權公開 (校外網路)

簡易檢索 / 詳目顯示

相關論文