Back to results list
Show full item record
Please use this identifier to cite or link to this item:
|Title:||Digital video browsing using efficient bitstream switching techniques||Authors:||Ip, Tak-piu||Degree:||M.Phil.||Issue Date:||2007||Abstract:||Recent advances in networking and multimedia technologies open the possibilities for network/service/content providers to offer residential customers with video-on-demand services. Interactive browsing operations such as random-access and fast-forward/backward playback are desirable features in these services. However, many video coding standards use motion-compensated prediction to reduce temporal redundancy of video sequences. Although it is efficient, this predictive scheme imposes extra constraints on how a compressed video displays since a predicted frame cannot be restored before any of its reference frames. It means that a compressed video should be played back in a pre-determined frame order. Displaying the digital video in other orders always requires extra resources for both network traffic and decoder complexity. Recently, a dual-bitstream technique has been suggested storing an additional reverse-encoded bitstream in the server to facilitate video browsing. Once the client requests a Video Cassette Recorder (VCR) operation, the server will select an appropriate frame for the client from either the forward or reverse-encoded bitstreams by considering the decoding effort at the decoder and the traffic over the networks. Therefore, bitstream switching is necessary for selecting appropriate frames in these two bitstreams. However, switching between these bitstreams is not a straightforward task. Since a P-frame is encoded using the prediction from the previously reconstructed reference, switching between the bitstreams at a P-frame would lead to drift errors due to the mismatch of the reconstructed references at that frame. Such errors will be propagated to subsequent P-frames. In this thesis, we investigate the impact of bitstream switching between the forward and reverse-encoded bitstreams. Two approaches based on both frame level and macroblock level are then proposed to solve the drift problem. In the proposed frame-level approach, we modify the original dual-bitstream structure by adopting Sl/SP-frames to eliminate drift errors and some efficient algorithms are investigated for implementing the required components and addressing this challenging issue. Experimental results show that, as compared to the original dual-bitstream structure, this new approach enhances the quality of the reconstructed video significantly. For the proposed macroblock-level approach, a video server classifies macroblocks in the requested frame into two categories - a reference-mismatched macroblock (RMMB) and a non-reference-mismatched macroblock (non-RMMB). A novel technique is used to manipulate the necessary macroblocks in the compressed domain and then the server sends the processed macroblocks to the client machine. For non-RMMBs, we propose a sign inversion technique in the Variable Length Coding (VLC) domain to eliminate the drift errors at certain areas that have static or slow motion activities. Besides, a simple version for low-cost video servers, which makes use of the redundancy inherent between the forward and reverse-encoded bitstreams in order to achieve a substantial reduction on the size of the reverse-encoded bitstream, is also proposed. It is exciting to report in this thesis that significant improvements in terms of the server complexity, the storage requirement of a server, and the browsing quality of reconstructed video can be achieved by employing our frame-level and macroblock-level approaches. Undoubtedly, these techniques are able to effectively facilitate video browsing in the dual-bitstream system.||Subjects:||Hong Kong Polytechnic University -- Dissertations.
Video dial tone.
|Pages:||xv, 150 leaves : ill. ; 30 cm.|
|Appears in Collections:||Thesis|
View full-text via https://theses.lib.polyu.edu.hk/handle/200/1373
Citations as of May 22, 2022
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.