- 學位論文
用於資料保存上的RVC視訊轉碼器及位元率控制方法
Reconfigurable Video Transcoder and Its Rate Control Methods for Long Term Data Preservation
摘要
近年來隨著多媒體科技越發進步、頻寬之不斷增加，以及網路的普及化，使得人們在資訊的取得日益方便。而網路及多媒體科技的急速發展也使得多媒體服務提供者能夠提供多樣化的網路多媒體服務，例如隨選視訊、遠距教學、及影片分享等。用戶可利用不同之多媒體接取裝置，例如個人電腦、3G行動電話及PDA等等，以多媒體串流技術很方便地擷取即時或預存之多媒體資訊。 在一般影片中，視訊的資料量極大，為有效壓縮資料及方便管理，目前視訊資料大多以MPEG-x、H.26x或其他商業公司自有的編碼格式（例如，Microsoft WMV-9或RealNetwork等等）壓縮並儲存。當需要觀賞視訊時，我們會需要相對應的編解碼器，進行視訊的編碼或解碼，甚或視訊轉換編碼(video transcoding)。可以預期的是，隨著編解碼技術的進步，未來將不斷會有新標準被提出，而開發人員便必須及時開發出相對應的編解碼器。這對於立即的使用需求及開發人力都是不經濟的。另一方面，在編解碼器中，類似的運算單元通常擁有相似的演算法，而新的編解碼標準相對於舊的，總是會有相當大比例的運算單元重複使用或是僅僅部分不同而已。假如我們可以把這些運算單元標準化，往後的新標準將能夠很輕易得使用現有的運算單元建構出來. 本研究將採用MPEG正在發展的Reconfigurable Video Coding (RVC) 標準所提供的動態地及彈性地建構解碼器的機制，發展Universal video transcoder的彈性化架構。首先將分析MPEG-2、MPEG-4編解碼標準的架構及功能單元(Functional Unit)，建立功能單元的程式庫。並將以所建立的程式庫為基礎實現MPEG-2轉MPEG-4視訊轉碼器，達成具有可擴充能力之視訊轉碼器架構。接著，我們也將在此RVC視訊轉碼器平台上，實現應用於轉碼器架構之最佳化位元率控制演算法，並希望能在固定影像品質的前題下，將視訊轉碼成最小位元率，使其更適合於資料保存的需求上。
並列摘要
In recent years, multimedia improving, bandwidth increasing and the popular for internet have made information receiving much more convenient. The fast growing of digital content technique keeps multimedia service giver provide more internet service, including video on demand, distance learning and movie sharing. User can take advantage of different multimedia access device (e.g. computer, 3G mobile, PDA, etc. ) to conveniently retrieve real-time data or pre-stored information by media streaming technique. In the digital archives, videos are the most amounts of data. In order to reduce the storage needs of video data, the videos have usually been compressed by the MPEG-x/H.26x codec or the proprietary codec (for example, Microsoft WMV-9 or RealNetwork). When we want to watch video, we need corresponding decoder/encoder to carry out decoding/encoding ; For video format transformation, we use transcoder to increase transform efficiency and to save time. With the development of video compression technology, high efficiency video coding would be used to compress video data. Therefore, the digital archives would contain various videos compressed by various video coding techniques. With the use of new video format, the programmer must to build corresponding codec timely. It is diseconomy for instant requirement and manpower. On the other hand, analogue operation functional unit usually owns similar algorithm for video codec. Compare late coding format to the old one, and we know most of the operation function are reused. If we can standardize these functional units, new coding tool will be assembled easily. In this research, we adopt the Reconfigurable Video Coding (RVC) technique by which a decoder can be built dynamically and flexibly, to develop the universal video transcoder. We will analyze the various functional units of the MPEG-2/4 video coding standards, and then implement the functional units. Based on the function units repository, we will construct the universal video transcoder in order to achieve the reconfigurable transcoder architecture. We would implement the optimized rate control algorithm for the RVC transcoding platform. Our research also discussed how to transcode the video bistream to the least bit rate in constant quality. The constant-quality variable bit rate algorithm is useful in data saving application.