A Study of H.264/AVC-Based Scalable Video Coding Using Computer Vision Techniques

View Statistics Email Alert RSS Feed

Information

Details

Project title

Code/計畫編號

NSC97-2221-E019-032

Translated Name/計畫中文名

結合電腦視覺技術之快速H.264/AVC為基礎的可調適視訊編碼研究

Project Coordinator/計畫主持人

Shyi-Chyi Cheng

Funding Organization/主管機關

National Science and Technology Council

Department/Unit

Department of Computer Science and Engineering

Website

https://www.grb.gov.tw/search/planDetail?id=1690510

Year

2008

Start date/計畫起

01-08-2008

Expected Completion/計畫迄

01-07-2009

Bugetid/研究經費

614千元

ResearchField/研究領域

資訊科學--軟體

"嶄新的視訊傳播系統創造了具彈性、可調適的視訊數位資料項的技術需求，不同型態的前端裝置如PC、PDA、多媒體終端機、智慧型手機、車用多媒體裝置等均已透過不同型態的網路(有線或無線)連結到各別的多媒體伺服器，異質化的視訊處理環境大大增加信號處理的複雜度。然而，配合無線網路的快速發展，無所不在的視訊傳播系統正一步一步成長中。可調式視訊編碼技術，是一項可解決當前化的視訊處理環境的信號處理問題的備受矚目的視訊傳播技術，本計畫討論SVC 的主要關鍵技術及其編碼效率，並提出一適合平行化處理、以軟體為主的快速SVC 編碼架構，據以改善兼顧速率-失真的SVC 的編碼效率，軟體式平行處理架構具有彈性的處理方式，更適合用以架設即時的SVC 編碼器；值得一提的是，本計劃所提出的軟體式SVC 平行處理架構，不僅可提供SVC 時間可調適、空間可調適、及畫質可調適之三種視訊調適服務，我們也設計一個即時SVC 平台調適架構，所謂的平台調適指的是我們所提出的軟體式 SVC 可應用於不同處理速度的硬體平台，即處理器計算速度也包含SVC 視訊調適服務的考慮範圍，直到目前為止，就筆者個人知識所及，仍未有相關文獻報導相關作法。可調適平台的SVC 編碼器可進一步擴展SVC 的潛在應用層面。本計畫為一三年期計畫，第一年，我們將設計一基於線邊對齊模式的快速模式決策，這個方法的基本概念在於使用電腦視覺技術，同時改善H.264/AVC 的編碼效率及提出符合人類視覺系統的編碼結果；第二年，根據第一年的研究成果，我們將提出一即時軟體式的H.264/AVC 平行處理架構，這個架構不僅可進一步提昇 H.264/AVC 的編碼效率，而且可提供平台調適性(Platform Scalability)，即在計算能力不同的工作平台上，進行即時的視訊編碼動作；第三年則整合前兩年的研究成果，擴展為包含時間可調適、空間可調適、畫質可調適、及平台可調適四維度的SVC 編碼系統，初步實驗結果驗證本計畫提出的方法的可行性。" "Scalable video coding (SVC) techniques meet the requirements of flexible video communication systems to transmit video data to a variety of multimedia terminals including PC, PDA, smart phones, and vehicle multimedia devices. Personalized multimedia services are essential for the state-of-the art Internet applications and the goals can be achieved by adapting multimedia content to users’ preferences and perceptual characteristics. As the rapid advance of wireless networks, the ubiquitous video transmission systems are evolving. SVC is a promising technique to process and transmit video data over heterogeneous video communication environments. In this project, we discuss the core technology of SVC and its coding efficiency in detail. A software-based real-time SVC architecture, which is suitable to implement in parallel is also proposed. The proposed SVC system offers spatial scalability, temporal scalability, quality scalability, and computation scalability based on a rate-distortion-computation optimization process. The so-called computation scalability is to provide the functionality of video coding in high quality even on a system with less computing power. To the knowledge of the authors, there is not any literature presenting the method to offer the computation scalability in a SVC system. The schedule of the project is divided into three years. In the first year, we will propose a fast motion estimation scheme for H.264/AVC based on line edge alignment between two image blocks. Line edge is an important feature to the human visual system and widely used in low-level computer vision. The coding results of the proposed fast H.264/AVC preserve important features of the human visual system. In the second year, we focus on the design of a computation-aware H.264/AVC using content-based image indexing techniques. The problem of large amount of memory accesses in H.264/AVC is properly solved by the indexing structure which offers effective random access capability to the video coding system. As compared with the hardware speedup methods, the proposed method offers much fine-grained parallelism which is more suitable to implement in parallel. The computation-aware characteristics of the method span a new way to design a real-time H.264/AVC system. Finally, in the last year, we aim at the design of a H.264/AVC-based SVC system which is basically an extension of proposed H.264/AVC system discussed in the last two years. The proposed SVC system offers spatial scalability, temporal scalability, quality scalability, and computation scalability based on a rate-distortion-computation optimization process. Preliminary experimental results show the feasibility of the proposed methods."

Keyword(s)

視訊傳播系統
可調式視訊編碼
平行處理
內嵌式系統
電腦視覺
速率-失真度-計算最佳化
H
264/AVC
Video communication
scalable video coding
parallel processing
embedded system
computer vision
rate-distortion-computation optimization
H
264/AVC

DSpace CRIS

A Study of H.264/AVC-Based Scalable Video Coding Using Computer Vision Techniques

Details

Description