Système unifié de transcodage vidéo permettant la réalisation d'une combinaison d'opérations de transcodage

Abstract

Over the years, the use of the video in our daily life has been increasing, thanks to the internet and the various sorts of mobile deviees: cellular telephones, persona} digital assistants (PDA), smartphones, etc. These deviees have already exceeded the number of persona} computers. The heterogeneity of networks (cable, wireless networks, optical fi bers, etc.), the variety of deviees as well as the various deployed multimedia applications have made indispensable the adaptation of the multimedia contents in order to facilitate universal and transparent access to the final users. Video transcoding is a key technology which allows mitigating this problem by transforming the previously encoded video according to the new transmission and/or user needs. Thus, transcoding has become a very active research area in which severa} transcoding architectures and systems have been proposed for various use cases such as bitrate adaptation, spatial resolution adaptation, temporal resolution adaptation, format adaptation, logo insertion, etc. However, the majorities of proposed solutions were standalone and have concentrated on the study of various video transcoding operations separately. In this project, we analyze the various video architectures presented in the literature and propose a unified architecture allowing the realization of various transcoding operations addressing a large number of use cases in a single system. Then, we propose variants to the existing video transcoding algorithms which are better adapted to our system. In order to evaluate the proposed unified video transcoding system, experiments are performed for different combinations of video transcoding operations. Our results show that the proposed system permits an average of 2.6 more channels with negligible quality Joss of 0.12 dB for the bitrate adaptation from 512 to 384 kb/s. For the spatial resolution adaptation from 512 to 256 kb/s, an average of2.14 more channels with 0.95 dB loss is obtained. For the format adaptation from 512 to 512 kb/s, the obtained results show an average of 1.84 times more channels with quality loss of 1.14 dB. Finally, for the format adaptation with spatial resolution adaptation from 256 to 32 kb/s, an average of 1.19 more channels with 0.41 dB loss is obtained.

Date	8 Sept 2010
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Stéphane Coulombe (Supervisor)