Elsevier Editorial System(tm) for Journal of Visual Communication and Image Representation Manuscript Draft Manuscript Number: Title: Efficient transcoding of an MPEG-2 bit stream to an H.264 bit stream Article Type: Regular Article Keywords: MPEG-2; H.264; video transcoding Corresponding Author: Mr. Att Kruafak, Sc.M.EE. Corresponding Author's Institution: The University of Texas at Arlington First Author: Rochelle Pereira, MS.EE. Order of Authors: Rochelle Pereira, MS.EE.; K. R Rao, PhD.; Att Kruafak, Sc.M.EE. Abstract: The objective of this paper is to develop a technique for transcoding from MPEG-2 main profile to H.264 main profile and compare its performance with other transcoding architectures. The proposed transcoder reuses information from the MPEG-2 bit stream talking into account the improved techniques such as multiple block size motion estimation, in loop deblocking filter, intra directional prediction, etc., adopted in H.264. The proposed method achieves low complexity, comparable quality and reduced bit rate in the transcoding process.Enclosed is the manuscript "Efficient transcoding of an MPEG-2 bit stream to an H.264 bit stream" for possible publication in Journal of Visual Communication and Image Representation. Please address all your correspondence to me. K. R. Rao, Electrical Engineering Department, The University of Texas at Arlington, 416 Yates Street, Box 19016, Arlington, Texas 76019, USA Phone: +1 817-272-3478, Fax: +1 817-272-2253, Email: [email protected] and Att Kruafak Electrical Engineering Department, The University of Texas at Arlington, 416 Yates Street, Box 19016, Arlington, Texas 76019, USA Phone: +1 682-553-5358, Fax: +1 817-272-2253, Email: [email protected] Cover LetterEfficient transcoding of an MPEG-2 bitstream to an H.264 bit streamRochelle Pereiraa,1,K.R.Raob,2and Att Kruafakb,3aNVIDIA Corp., Santa Clara, CA 95050, USAbElectrical Engineering Department, The U niversity of Texas at Arlington,416 Yates Street, Box 19016, Arlington, Texas 76019, USA.AbstractThe objective of this paper is to develop a technique for transcoding from MPEG-2main profile to H.264 main profile and compare its performance with other transcod-ing architectures. The proposed transcoder reuses information from the MPEG-2 bitstream talking into account the improved techniques such as multiple block size mo-tion estimation, in loop deblocking filter, intra directional prediction, etc., adoptedin H.264. The proposed method achieves low complexity, comparable quality andreduced bit rate in the transcoding process.Key words: MPEG-2, H.264, video transcodingPACS:1 IntroductionMPEG-2 [1,2] has been a widely accepted video coding standard for variousapplications ranging from DVD to digital TV broadcast. A large variety ofproducts based on the MPEG-2 standard are available in the market. Themost important goal of MPEG-2 is to make the storage and transmissionof digital AV material more efficient. The latest standard, H.264/AVC [3–5],has an even broader perspective to support high and low bit rate multimediaapplications on existing and future networks. The advantage in terms of betterquality at a lower bit-rate is why H.264 is fast replacing MPEG-2 [6]. However,1Email address: [email protected] address: [email protected], Phone: +1-817-272-3478, Fax: +1-817-272-2253,URL: http://www-ee.uta.edu/dip3Email address: [email protected] submitted to Elsevier Science 23 February 2007* Manuscriptthe user end hardware such as set-top-boxes, DVD players had previously beenadapted for MPEG-2 coded bit streams. This gives rise to a need for portabilitybetween MPEG-2 and H.264.Video transcoding is the operation of converting video from one format toanother [7–10]. A format is defined by characteristics such as bit-rate, spatialresolution, etc. One of the earliest applications of transcoding is to adaptthe bit-rate of a compressed stream to the channel bandwidth for universalmultimedia access in various channels like wireless networks, Internet, dial-up networks, etc. Changes in the characteristics of an encoded bit streamlike bit-rate, spatial resolution, quality, etc., can also be achieved by scalablevideo coding [8]. However, in cases where the available network bandwidth isinsufficient or if it fluctuates with time, it may be difficult to set the base layerbit-rate. In addition, scalable video coding demands additional complexitiesat both the encoder and the decoder.2 Transcoding ArchitecturesThe basic architecture for converting an MPEG-2 elementary stream into anH.264 elementary stream arises from complete decoding of the MPEG streamand then re-encoding into an H.264 stream. However, this involves significantcomputational complexity [7]. Hence there also is a need to transcode at lowcomplexity. Transcoding can in general be implemented in the spatial domainor in the transform domain or in a combination of the two domains.Several transcoding architectures have been proposed earlier, i.e.,2.1 Open Loop Transform Domain Transcoding (Fig. 1) [8]Open loop transcoders are computationally efficient. They operate in the DCTdomain. However they are subject to drift error. Drift error occurs due torounding, quantization loss and clipping functions.2.2 Cascaded Pixel Domain Architecture (CPDT) (Fig. 2) [8]This is the most basic transcoding architecture. The motion vectors from theincoming bit stream are extracted and reused. Thus the complexity of themotion estimation block is eliminated which accounts for 60 % of the encodercomputation [11]. As compared to the previous architecture, CPDT is drift2Fig. 1. Open loop transform domain transcoder architecture.Fig. 2. Cascaded pixel domain transcoder architecture.free. Hence, even though it is slightly more complex, it is suitable for heteroge-neous transcoding among different standards where the basic parameters likemode decisions, motion vectors, etc., are to be re-derived.3Fig. 3. Simplified transform domain transcoder architecture.Fig. 4. Transform domain motion compensation illustration.2.3 Simplified DCT Domain Transcoders (SDDT) (Fig. 3) [8]This transcoder is based on the assumption that DCT, IDCT and motion com-pensation are linear processes. This architecture requires that motion compen-sation be performed in the DCT domain, which is a major computationallyintensive operation [12]. For instance, as shown in Fig. 4, the goal is trying tocompute the DCT coefficients of the target
View Full Document
Unlocking...