An Efficient Motion Estimation Method for H.264-Based Video Transcoding with Arbitrary Spatial Resolution Conversion

H.264;video transcoding;Electrical and Computer Engineering

Wang, Jiao

University of Waterloo

关键词: H.264;    video transcoding;    Electrical and Computer Engineering;   

Others  :  https://uwspace.uwaterloo.ca/bitstream/10012/3133/1/JiaoWang_thesis.pdf

瑞士|英语

来源: UWSPACE Waterloo Institutional Repository

【 摘 要 】

As wireless and wired network connectivity is rapidly expandingand the number of network users is steadily increasing, it has become moreand more important to support universal access of multimediacontent over the whole network. A big challenge, however, isthe great diversity of network devices from full screen computersto small smart phones. This leads to research on transcoding,which involves in efficiently reformatting compressed data fromits original high resolution to a desired spatial resolutionsupported by the displaying device. Particularly, there is agreat momentum in the multimedia industry for H.264-basedtranscoding as H.264 has been widely employed as a mandatoryplayer feature in applications ranging from television broadcastto video for mobile devices.While H.264contains many new features for effective videocoding with excellent rate distortion (RD) performance, a major issuefor transcoding H.264 compressed video from one spatial resolutionto another is the computational complexity. Specifically, it isthe motion compensated prediction (MCP) part. MCP is the maincontributor to the excellent RD performanceof H.264 video compression, yet it is very time consuming. In general,a brute-force search is used to find the best motion vectors for MCP.In the scenario of transcoding, however, an immediate idea forimproving the MCP efficiency for the re-encoding procedure is toutilize the motion vectors in the original compressed stream.Intuitively, motion in the high resolution scene is highly relatedto that in the down-scaled scene.In this thesis, we study homogeneous video transcoding from H.264to H.264. Specifically, for the video transcoding with arbitraryspatial resolution conversion, we propose a motion vector estimationalgorithm based on a multiple linear regression model, whichsystematically utilizes the motion information in the original scenes.We also propose a practical solution for efficiently determining areference frame to take the advantage of the new feature of multiplereferences in H.264. The performance of the algorithm was assessedin an H.264 transcoder. Experimental results show that, as comparedwith a benchmark solution, the proposed method significantly reducesthe transcoding complexity without degrading much the video quality.